Discrimination system and method

The system uses specular and diffusely reflected light amounts, basis weight, and thickness or density values with machine learning to generate a trained model for paper type discrimination, addressing the limitation of existing systems in identifying unregistered paper types, thereby enhancing the accuracy and adaptability of paper type identification.

JP7589843B2Active Publication Date: 2024-11-26KONICA MINOLTA INC
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
JP2024008648
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2024-01-24
Publication Date
2024-11-26
Estimated Expiration
2039-10-09

AI Technical Summary

Technical Problem

Existing paper type discrimination systems, such as those described in Patent Document 1, are unable to accurately determine the type of paper that is not registered in their database.

Method used

A system and method utilizing specular and diffusely reflected light amounts, basis weight, and thickness or density values, combined with machine learning, to generate a trained model for paper type discrimination, enabling identification of unknown paper types.

Benefits of technology

Enables accurate determination of paper types not previously registered in the database, improving the adaptability and accuracy of paper type identification in image forming apparatuses.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007589843000004
    Figure 0007589843000004
  • Figure 0007589843000005
    Figure 0007589843000005
  • Figure 0007589843000006
    Figure 0007589843000006
Patent Text Reader

Abstract

To provide a discriminating device capable of appropriately discriminating a paper type even for a kind of paper not previously registered in a paper type database.SOLUTION: A discriminating device has an acquiring section and a discriminating section. The acquiring section acquires a specular reflection light amount value relating to an amount of specular reflection light specularly reflected by a surface of a recording medium when the recording medium is irradiated with light from a light source via an optical system at a predetermined angle of incidence, a diffused reflection light amount value relating to an amount of diffused reflection light diffusely reflected at least at one reflection angle on the surface of the recording medium, and a value relating to a basis weight of the recording medium and a value relating to a thickness or density of the recording medium. The discriminating section inputs the respective values acquired by the acquiring section into a machine-learned trained model for paper type discrimination to discriminate the type of the recording medium.SELECTED DRAWING: Figure 21
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a discrimination system and a discrimination method. [Background technology]

[0002] In recent years, image forming devices such as electrophotographic printers have come to be widely used in the color printing industry. In the field of PP (production print) corresponding to the color printing industry, there is a demand for adaptability to a wider variety of papers than those used in offices. In order to perform high-quality printing on these various papers, there are image forming devices that set multiple items for the characteristics of the papers stored in the paper feed tray and print under image formation conditions according to the items that have been set.

[0003] In order to set such a variety of papers, there are image forming devices equipped with sensors that automatically detect the characteristics of the paper used for printing. For example, Patent Document 1 proposes that an arithmetic formula for calculating smoothness from a plurality of detection output amounts (signal strengths) that detect the physical property values ​​of the paper is calculated in advance as a linear regression formula, and the smoothness is calculated using the arithmetic formula to determine the paper type. In addition, the above technology finds and displays candidates for paper types (brands) that have the closest physical property values ​​and smoothness values ​​to the detected paper physical property values ​​and registered physical property values ​​and smoothness values ​​from a paper type database (paper profile) that is stored in advance in the image forming device. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-223692 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, the technique of Patent Document 1 has a problem in that it is not possible to determine a candidate paper type (brand) for new paper that is not registered in the paper type database stored in advance.

[0006] The present invention has been made in consideration of the above circumstances, and aims to provide a paper type discrimination device, a paper type discrimination system, and a paper type discrimination method that can properly discriminate the paper type even for paper types that have not been registered in advance. [Means for solving the problem]

[0007] The above object of the present invention can be achieved by the following means.

[0008] (1A) an acquisition unit that acquires a specular reflection light quantity value relating to the quantity of light that is specularly reflected from a surface of the recording medium when light from a light source is irradiated onto a recording medium via an optical system at a predetermined incident angle, a diffuse reflection light quantity value relating to the quantity of light that is diffusely reflected at at least one reflection angle from the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; A learning unit that generates a trained model for paper type discrimination by performing machine learning; A discrimination unit that inputs each value acquired by the acquisition unit into the machine-learned model to discriminate the type of the recording medium; an output unit that outputs a discrimination result by the discrimination unit; a reception unit that receives an instruction from a user to change the determination result; A storage unit that stores teacher data for executing the machine learning, The learning unit uses change information, including the changed discrimination result and information input into the trained model to obtain the discrimination result, as the teacher data to further machine-learn the trained model.

[0009] (1B) an acquisition unit that acquires a specular reflection light quantity value relating to the quantity of light that is specularly reflected from a surface of the recording medium when light from a light source is irradiated onto the recording medium via an optical system at a predetermined incident angle, a diffuse reflection light quantity value relating to the quantity of light that is diffusely reflected at at least one reflection angle on the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; A learning unit that generates a trained model for paper type discrimination by performing machine learning; A discrimination unit that inputs each value acquired by the acquisition unit into the machine-learned model to discriminate the type of the recording medium. A discrimination system in which the learning unit performs machine learning at a preset timing.

[0010] (1C) an acquisition unit that acquires a specular reflection light quantity value relating to the quantity of light that is specularly reflected from a surface of the recording medium when light from a light source is irradiated onto the recording medium via an optical system at a predetermined incident angle, a diffuse reflection light quantity value relating to the quantity of light that is diffusely reflected at at least one reflection angle on the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; A learning unit that generates a trained model for paper type discrimination by performing machine learning; A discrimination unit that inputs each value acquired by the acquisition unit into the machine-learned model to discriminate the type of the recording medium. the learning unit is provided in a server connected to the image forming apparatus via a network, the acquiring unit and the determining unit are provided in the image forming apparatus, The server further trains the learning unit to machine-learn the trained model, and transmits the machine-learned trained model to the image forming device; The discrimination unit of the image forming apparatus is a discrimination system that discriminates the type of the recording medium using the trained model transmitted from the server.

[0011] (1D) an acquisition unit that acquires a specular reflection light quantity value relating to the quantity of light specularly reflected light that is irradiated from a light source via an optical system onto a recording medium at a predetermined incident angle and specularly reflected on a surface of the recording medium, a diffuse reflection light quantity value relating to the quantity of light diffusely reflected at at least one reflection angle on the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; A learning unit that generates a trained model for paper type discrimination by performing machine learning; A discrimination unit that inputs each value acquired by the acquisition unit into the machine-learned model to discriminate the type of the recording medium. the learning unit and the determination unit are provided in a server connected to the image forming apparatus via a network, the acquisition unit is provided in the image forming apparatus, the image forming apparatus transmits each value acquired by the acquisition unit to the server; The discrimination unit of the server is a discrimination system that discriminates the type of the recording medium using the values ​​transmitted from the image forming device and the trained model.

[0012] (1) A discrimination device having an acquisition unit that acquires a specular reflection light quantity value relating to the amount of light that is specularly reflected from a surface of the recording medium when the light source is irradiated onto the recording medium via an optical system at a predetermined incident angle, a diffuse reflection light quantity value relating to the amount of light that is diffusely reflected from the surface of the recording medium at at least one reflection angle, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium, and a discrimination unit that inputs each of the values ​​acquired by the acquisition unit into a trained model for paper type discrimination that has been machine-learned to discriminate the type of the recording medium.

[0013] (2) The discrimination device described in (1) above, wherein the trained model is composed of a random forest.

[0014] (3) The discrimination device according to (1) above, wherein the trained model is constituted by a neural network.

[0015] (4) A discrimination system having the discrimination device described in any one of (1) to (3) above, and a learning unit that generates the trained model by performing machine learning using the specular reflection light value, the diffuse reflection light value, and a value related to the thickness and basis weight of the recording medium or a value related to the density of the recording medium as input factors of training data, and the type of the recording medium as an output factor of the training data.

[0016] (5) The discrimination system according to (4) above, further comprising an image forming device for forming an image on the recording medium.

[0017] (6) The image forming device is a discrimination system described in (5) above, which has a light source, an optical system that irradiates light from the light source onto the surface of a recording medium in an irradiation area at a predetermined incident angle, a first light receiving unit that detects the amount of light of specularly reflected light that is specularly reflected on the surface of the recording medium in the irradiation area, at least one second light receiving unit that detects the amount of light of diffusely reflected light that is diffusely reflected at at least one reflection angle on the surface of the recording medium in the irradiation area, and a detection unit that detects a value related to the basis weight of the recording medium and a value related to the thickness or density of the recording medium.

[0018] (7) The discrimination system according to (5) or (6) above, wherein the learning unit performs machine learning at a time when image formation processing is not being performed in the image forming device.

[0019] (8) A discrimination system described in any of (5) to (7) above, wherein the control unit of the image forming device does not accept instructions to execute an image forming process while machine learning is being performed in the learning unit.

[0020] (9) The discrimination system according to any one of (5) to (8) above, wherein the learning unit performs machine learning at a preset timing.

[0021] (10) A discrimination system described in any of (5) to (9) above, further comprising an output unit that outputs the discrimination result by the discrimination unit, a reception unit that receives instructions from a user to change the discrimination result, and a memory unit that stores the teacher data for performing the machine learning, wherein the learning unit uses change information including the changed discrimination result and information input into the trained model to obtain the discrimination result as the teacher data to further machine learn the trained model.

[0022] (11) The memory unit and the learning unit are provided in a server connected to the image forming device via a network, and the acquisition unit, the discrimination unit, the output unit, and the reception unit are provided in the image forming device, and the image forming device transmits the change information received by the reception unit to the server, and the server stores the change information sent from the image forming device as the teacher data in the memory unit, further trains the learning unit to machine learn the trained model using the stored teacher data, and transmits the machine-learned trained model to the image forming device, and the discrimination unit of the image forming device discriminates the type of the recording medium using the trained model sent from the server.

[0023] (12) The discrimination system described in (10) above, wherein the memory unit, the learning unit, and the discrimination unit are provided in a server connected to the image forming device via a network, the acquisition unit, the output unit, and the reception unit are provided in the image forming device, the image forming device transmits each value acquired by the acquisition unit to the server, and the discrimination unit of the server discriminates the type of the recording medium using each value transmitted from the image forming device and the trained model.

[0024] (13) A discrimination system as described in (11) or (12) above, wherein the change information transmitted from the image forming device is stored in the memory unit as the training data, and the learning unit further performs machine learning on the trained model using the stored training data based on predetermined execution conditions.

[0025] (14A) a step (a) of acquiring a specular reflection light quantity value relating to the quantity of light specularly reflected from a surface of the recording medium, the specular reflection light quantity value relating to the quantity of light diffusely reflected from the surface of the recording medium at at least one reflection angle, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; (b) generating a trained model for paper type discrimination by performing machine learning using the training data; A step (c) of inputting each value acquired in the step (a) into the trained model trained by machine learning in the step (b) to determine the type of the recording medium; a step (d) of outputting a result of the discrimination by the step (c); (e) receiving an instruction from a user to change the determination result; having In the step (b), modification information including the modified discrimination result and information input into the trained model to obtain the discrimination result is used as the training data, thereby further machine learning the trained model.

[0026] (14B) a step (a) of acquiring a specular reflection light quantity value relating to the quantity of light specularly reflected from a surface of the recording medium, the specular reflection light quantity value relating to the quantity of light diffusely reflected at at least one reflection angle from the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium; (b) generating a trained model for paper type discrimination by performing machine learning; and (c) inputting each value acquired in the step (a) into the trained model trained in the step (b) to determine the type of the recording medium, A discrimination method, wherein in the step (b), machine learning is performed at a preset timing.

[0027] (14C) A step (a) of acquiring a specular reflection light quantity value relating to the quantity of light specularly reflected from a surface of the recording medium, the specular reflection light quantity value relating to the quantity of light diffusely reflected from the surface of the recording medium at at least one reflection angle, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium, (b) generating a trained model for paper type discrimination by performing machine learning; and (c) inputting each value acquired in the step (a) into the trained model trained in the step (b) to determine the type of the recording medium, The step (b) is executed by a server connected to the image forming apparatus via a network, the step (a) and the step (c) are performed by the image forming apparatus; In the step (b), the server further trains the trained model by machine learning and transmits the trained model to the image forming device; A discrimination method, in which in the step (c), the image forming device discriminates the type of the recording medium using the trained model transmitted from the server.

[0028] (14D) a step (a) of acquiring a specular reflection light quantity value relating to the quantity of light specularly reflected from a surface of the recording medium, the specular reflection light quantity value relating to the quantity of light diffusely reflected at at least one reflection angle from the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium, the step (a) of acquiring a specular reflection light quantity value relating to the quantity of light diffusely reflected from the surface of the recording medium at a predetermined incident angle from a light source through an optical system, the step (a) of acquiring a specular reflection light quantity value relating to the quantity of light diffusely reflected from the surface of the recording medium at at least one reflection angle, the value relating to the basis weight of the recording medium, and the value relating to the thickness or density of the recording medium; (b) generating a trained model for paper type discrimination by performing machine learning; and (c) inputting each value acquired in the step (a) into the trained model trained in the step (b) to determine the type of the recording medium, the step (b) and the step (c) are executed by a server connected to the image forming apparatus via a network; The step (a) is performed in the image forming apparatus, The image forming apparatus transmits each value acquired in the step (a) to the server; A discrimination method in which, in step (c), the server discriminates the type of the recording medium using the values ​​transmitted from the image forming device and the trained model.

[0029] (14) A specular reflection light quantity value relating to the quantity of light that is irradiated from a light source through an optical system onto a recording medium at a predetermined incident angle and specularly reflected by the surface of the recording medium, and a diffuse reflection light quantity value relating to the quantity of light that is diffusely reflected at at least one reflection angle on the surface of the recording medium. A method for identifying a type of recording medium, comprising: a step (a) of acquiring a value relating to the basis weight of the recording medium and a value relating to the thickness or density of the recording medium; and a step (b) of inputting the values ​​acquired in step (a) into a machine-learned trained model for paper type identification to identify the type of the recording medium. Effect of the Invention

[0030] According to the discrimination device of the present invention, a specular reflection light amount value relating to the amount of specular reflection light reflected from the surface of the recording medium, a diffuse reflection light amount value relating to the amount of diffuse reflection light diffusely reflected at at least one reflection angle from the surface of the recording medium, a value relating to the basis weight of the recording medium, and a value relating to the thickness or density of the recording medium are obtained, and the obtained values ​​are input into a machine-learned trained model for paper type discrimination to discriminate the type of recording medium. This makes it possible to appropriately discriminate the paper type even for types of paper that are not registered in advance in the database. [Brief description of the drawings]

[0031] [Figure 1] 1 is a diagram showing a schematic configuration of an image forming system including an image forming apparatus according to an embodiment of the present invention. [Diagram 2] 11 is a side view showing the configuration of a media sensor disposed in a transport path. FIG. [Diagram 3] FIG. 2 is a perspective view showing the configuration of a basis weight sensor and a surface property sensor. [Figure 4] FIG. 2 is a cross-sectional view of a surface sensor. [Diagram 5] FIG. 2 is a perspective view showing an internal configuration of a surface sensor. [Figure 6] FIG. 2 is a schematic cross-sectional view of the surface sensor showing a state in which the shutter is open. [Figure 7] 3 is a schematic diagram showing the arrangement positions of a light-emitting unit and a light-receiving unit. FIG. [Figure 8] 4 is a schematic cross-sectional view showing the incident angle of irradiated light and the arrangement angle of a light receiving portion. FIG. [Figure 9] FIG. 2 is a schematic diagram showing the state of pulp fibers on the surface of paper. [Figure 10] FIG. 2 is a schematic diagram showing the microscopic variation in fiber orientation angle of paper. [Figure 11] 4A and 4B are schematic diagrams showing the surface distribution of a sheet of paper and the diffusion state of irradiated light. [Figure 12] 3 is a schematic diagram showing the irradiation diameter of irradiation light and the arrangement position of a light emitting unit (light source). FIG. [Figure 13] 11 is a graph showing variations in the amount of detected light due to differences in the irradiation diameter when multiple points are measured on the surface of a sheet. [Figure 14] 11 is a graph showing a change in detection accuracy when an error is added to input data. [Figure 15A] 11 is a graph showing the robustness of the mounting accuracy of the light-emitting unit. [Figure 15B] 11 is a graph showing the robustness of the mounting accuracy of the light-emitting unit. [Figure 16] 1 is a table showing the degree of correlation between the wavelength of light from a light-emitting portion and Beck smoothness. [Figure 17]1 is a graph showing a comparison between the Beck smoothness for four types of paper (gloss coated paper, matte coated paper, plain paper, and fine paper) and the predicted value based on the measurement results. [Figure 18A] 1 is a block diagram showing a configuration of an image forming apparatus; [Figure 18B] 2 is a block diagram showing a functional configuration of an overall control unit of the image forming apparatus according to the first embodiment; FIG. [Figure 19] 4 is a flowchart showing a printing process of the image forming apparatus. [Figure 20] 13 is a subroutine flowchart showing a paper setting process (step S10). [Figure 21] FIG. 13 is a control block diagram showing the determination process (S107). [Figure 22] FIG. 11 is a diagram for explaining a procedure of a paper setting process. [Figure 23] 13 is an example of an operation screen for receiving an instruction to start a determination process. [Figure 24] 13 is an example of an operation screen for receiving an instruction to execute a determination process. [Diagram 25] 13 is an example of an operation screen showing a determination result (paper type / basis weight classification). [Figure 26] 13 is an example of an operation screen showing a determination result (registered profile). [Figure 27] 13 is an example of an operation screen for receiving an instruction to start a process of changing the determined paper settings. [Figure 28] 13 is an example of an operation screen for receiving an instruction to change the determined paper setting. [Figure 29] 13 is an example of an operation screen for receiving an instruction to register the determined paper settings. [Diagram 30] FIG. 2 is a diagram illustrating an example of information registered in a database of teacher data. [Diagram 31] FIG. 1 is a diagram showing an example of a trained random forest model generated by machine learning using training data. [Diagram 32] 13 is a block diagram showing a functional configuration of an overall control unit of an image forming apparatus according to a second embodiment. FIG. [Diagram 33]13 is a flowchart showing a process of adding modification information to teacher data, which is executed in an image forming apparatus according to a second embodiment. [Diagram 34] 10 is a flowchart showing a process for generating and updating a paper type discrimination model and a paper type discrimination algorithm executed in an image forming apparatus according to a second embodiment. [Diagram 35] 11 is a diagram showing a procedure for generating a paper type discrimination model and converting the paper type discrimination model into a binary file. FIG. [Figure 36A] 13 is an example of a screen for accepting a setting of an execution condition for updating the paper type discrimination model. [Figure 36B] 13 is an example of a screen that accepts selection of a paper type discrimination model. [Figure 37] FIG. 2 is a block diagram showing the configuration of a server. [Figure 38] FIG. 3-1 is a block diagram showing the functional configuration of a control unit of the server according to the embodiment. [Figure 39] FIG. 3 is a block diagram showing a functional configuration of an overall control unit of an image forming apparatus according to a third embodiment. [Diagram 40] 13 is a flowchart showing a modification information transmission process executed in the image forming apparatus according to the third embodiment; [Diagram 41] 13 is a flowchart showing the teacher data adding process executed in the server according to the third embodiment. [Diagram 42] 13 is a flowchart showing a process for generating a paper type discrimination model and a paper type discrimination algorithm executed in the server according to the third embodiment; [Diagram 43] 13 is a flowchart showing a process of updating a paper type discrimination algorithm executed in an image forming apparatus according to a third embodiment; [Figure 44A] FIG. 3-2 is a block diagram showing the functional configuration of a control unit of the server according to the embodiment. [Figure 44B] FIG. 3-2 is a block diagram showing a functional configuration of an overall control unit of an image forming apparatus according to a third embodiment. [Diagram 45]20 is a flowchart showing a paper setting process (corresponding to step S10 in FIG. 19) executed in the image forming apparatus according to the third embodiment. [Figure 46] 13 is a flowchart showing a process of determining the paper type and the like executed in the server according to the third embodiment; [Figure 47] 13 is a flowchart showing a registration process of change information executed in the server according to the third embodiment. [Figure 48A] FIG. 3-3 is a block diagram showing the functional configuration of a control unit of the server according to the embodiment. [Figure 48B] FIG. 3-3 is a block diagram showing a functional configuration of an overall control unit of an image forming apparatus according to a third embodiment. [Figure 49] 13 is a flowchart showing a registration process of change information executed in the server according to the third embodiment; [Figure 50] 3-3 is a flowchart showing a generation and update process of a paper type discrimination model and a paper type discrimination algorithm executed in the server according to the embodiment; [Figure 51] FIG. 13 is a diagram showing an example of discriminating paper type using a trained model of a neural network that has been machine-learned using training data. [Figure 52] FIG. 2 is a diagram illustrating an example of an imaging mechanism for acquiring a surface image. [Figure 53] 1A to 1C are diagrams showing examples of surface images and characteristics of gloss coated paper, matte coated paper, fine paper, and plain paper captured by an imaging mechanism. [Figure 54] FIG. 2 is a schematic diagram showing a configuration of a basis weight sensor. [Figure 55] 4 is a block diagram showing the functional configuration of a basis weight sensor and a control unit. FIG. [Figure 56] FIG. 4 is a diagram showing a criterion representing the relationship between a first transmittance and a nominal basis weight. [Figure 57] FIG. 13 is a diagram showing a criterion representing the relationship between the second transmittance and the nominal basis weight. [Figure 58] FIG. 1 illustrates the correlation between transmittance at wavelength and nominal basis weight. [Figure 59] FIG. 13 is a diagram showing an index representing the correspondence relationship between transmittance and basis weight difference, and a basis weight threshold value. [Figure 60] FIG. 11 is a diagram showing an example of paper profile information. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0032] Hereinafter, an embodiment of the present invention will be described with reference to the attached drawings. In the description of the drawings, the same elements are given the same reference numerals, and duplicated description will be omitted. In addition, the dimensional ratios of the drawings are exaggerated for convenience of description, and may differ from the actual ratios. In the drawings, the up-down direction is the Z direction, the front and back directions of the image forming apparatus are the X direction, and the direction perpendicular to these X and Z directions is the Y direction. The X direction is also called the width direction or the rotation axis direction. In addition, in the periphery of the media sensor (media sensor 15 described later), the conveying direction of the recording medium that is parallel to the surface of the conveying path (conveying path 143 described later) that is inclined with respect to the horizontal plane and perpendicular to the X direction is called the Y' direction, and the direction perpendicular to this is called the Z' direction (see FIG. 2, etc.). In this embodiment, the recording medium includes printing paper (hereinafter simply referred to as paper) and various films. In particular, the paper includes those manufactured using mechanical pulp and / or chemical pulp derived from plants. The types of recording media include coated gloss paper and matte paper (gloss coated paper and matte coated paper), and uncoated plain paper and fine paper.

[0033] (First embodiment) Fig. 1 is a diagram showing a schematic configuration of an image forming system 1 including an image forming apparatus 10 according to the present embodiment. As shown in Fig. 1, the image forming system 1 includes the image forming apparatus 10, a paper feeder 20, and a post-processing apparatus 30, which are mechanically and electrically connected to each other.

[0034] (Image forming apparatus 10) The image forming apparatus 10 includes a control unit 11, a storage unit 12, an image forming unit 13, a paper feed conveyance unit 14, a media sensor 15, an operation panel 18, a communication unit (not shown), and the like. These are connected to each other via a signal line such as a bus for exchanging signals. FIG. 2 is a side view showing the configuration of the media sensor 15 arranged on the conveyance path 143. The media sensor 15 is composed of a paper thickness sensor 40, a basis weight sensor 50, a surface property sensor 60, and a paper pressing unit 70, and measures paper characteristics. This surface property sensor 60 functions as an optical sensor device and detects paper characteristics, particularly the surface property of the paper. The media sensor 15 including the surface property sensor 60 will be described in detail later. In this embodiment, the media sensor 15 functions as a detection unit. Moreover, the control unit 11 functions as a discrimination device. Furthermore, the control unit 11 functions as an output unit and a reception unit.

[0035] (Control unit 11) The control unit 11 is composed of a CPU, ROM, RAM, etc., and executes various processes by executing programs stored in the ROM or the memory unit 12 described below, and controls each part of the device and performs various arithmetic processes in accordance with the programs.

[0036] (Storage unit 12) The storage unit 12 is made up of auxiliary storage units such as a ROM for storing various programs and data in advance, a RAM for temporarily storing programs and data as a working area, and a hard disk for storing various programs and data. The storage unit 12 also stores information on paper stored in each paper feed tray. The paper information includes information on the brand, size (paper width, paper length), basis weight, and paper type (gloss coated paper, matte coated paper, plain paper, fine paper, rough paper, etc.) of the paper, which is set by a paper type determination process described later. The storage unit 12 may also store a trained model used to determine the paper brand or paper type, and a paper profile (both of which are described later).

[0037] (Image forming unit 13) The image forming unit 13 forms an image, for example, by an electrophotographic method. The image forming unit 13 includes writing units 131 corresponding to the basic colors of Y (yellow), M (magenta), C (cyan), and K (black), a photoconductor drum 132, and a developing unit 133 that contains a two-component developer consisting of a toner and a carrier of each color. The image forming unit 13 further includes an intermediate transfer belt 134, a secondary transfer unit 135, and a fixing unit 136. The toner images formed on the photoconductor drum 132 by the developing units 133 of each color are superimposed on the intermediate transfer belt 134, and are transferred to the paper S conveyed in the secondary transfer unit 135. The toner image on the paper S is fixed onto the paper S by being heated and pressed in the fixing unit 136 on the downstream side.

[0038] (Paper feed conveying section 14) The paper feed conveying unit 14 includes a plurality of paper feed trays 141, 142, conveying paths 143, 144, etc. The conveying paths 143, 144 include a plurality of conveying roller pairs provided along these conveying paths, and a drive motor (not shown) that drives these conveying roller pairs. A feed roller is provided to feed the topmost sheet of paper S among the plurality of sheets of paper S stacked and placed in the paper feed trays 141, 142, and sends the sheets S in the paper feed trays one by one to the conveying path on the downstream side. A media sensor 15 is disposed upstream of the registration rollers on the conveying path 143. As shown in FIG. 2, near the media sensor 15, the conveying path 143 includes an upper guide 1431 and a lower guide 1432 made of sheet metal, and the paper S passes between these guides that face each other at a predetermined interval.

[0039] The paper feed conveying section 14 conveys the paper S fed from the paper feed tray 141 or the like. The paper S conveyed along the conveying path 143 has an image formed on it by the image forming section 13, and then passes through the subsequent post-processing device 30 and is discharged onto the paper output tray 342. When performing double-sided printing in which an image is also formed on the back side of the paper S, the paper S with an image formed on one side is conveyed to a conveying path 144 for double-sided image formation located at the bottom of the device main body. The paper S conveyed to this conveying path 144 is turned over on a switchback path, and then merges with the conveying path 143 for one side, where an image is again formed on the other side of the paper S by the image forming section 13.

[0040] (Operation panel 18) The operation panel 18 is equipped with a touch panel, a numeric keypad, a start button, a stop button, etc., and is used to display the status of the image forming apparatus 10 or the image forming system 1 and to allow the user to set the type of paper placed in the paper feed tray 141, etc., and to input instructions.

[0041] (Paper feeder 20) 1, the paper feeder 20 includes a paper feed conveying section 24. In addition to the paper feed conveying section 24, the paper feeder 20 also includes a control section, a storage section, and a communication section (none of which are shown), which are connected to each other via signal lines such as a bus for exchanging signals. The paper feed conveying section 24 includes a plurality of paper feed trays 241, 242, 243, and a conveying path 244. The paper S conveyed from each paper feed tray is conveyed to the downstream image forming device 10, where the paper characteristics are measured by the media sensor 15 and an image is formed by the image forming section 13.

[0042] (Post-processing device 30) 1, post-processing device 30 includes post-processing section 31, conveying path 341, and paper discharge tray 342. Post-processing section 31 performs processes such as stapling, cutting, and punching (punching holes) on paper S conveyed from image forming device 10. In addition to these components, post-processing device 30 also includes a control section, a storage section, and a communication section (none of which are shown), which are connected to each other via signal lines such as a bus for exchanging signals.

[0043] (Media Sensor 15) As described above, the media sensor 15 is composed of the paper thickness sensor 40, the basis weight sensor 50, the surface property sensor 60, and the paper pressing unit 70. Referring to FIG. 2, of these components, the paper thickness sensor 40 is disposed on the most upstream side in the conveying direction. The paper thickness sensor 40 is composed of a pair of conveying rollers 411, 412, and a pressing mechanism 42 composed of an actuator, an encoder, and a light emitter / receiver. The upper conveying roller 411 is urged toward the lower conveying roller 412 by the pressing mechanism 42. As the paper S is conveyed to the nip between the conveying rollers 411, 412, the conveying roller 411 moves upward by a height corresponding to the thickness of the paper S. The pressing mechanism 42 detects the thickness of the paper S based on the amount of displacement of the conveying roller 411 in the height direction (thickness direction).

[0044] Fig. 3 is a perspective view showing the configurations of the basis weight sensor 50 and the surface property sensor 60. As shown in Fig. 2 and Fig. 3, the basis weight sensor 50 and the surface property sensor 60 are arranged side by side along the X direction (width direction) between a pair of transport rollers 1433 and 1434.

[0045] Basis weight sensor 50 is a sensor that detects the basis weight of a sheet, and includes a light-emitting section and a light-receiving section, and measures the basis weight based on the attenuation of light that passes through sheet S. For example, basis weight sensor 50 has a light-emitting section (see FIG. 54: first light-emitting section 51a, second light-emitting section 51b) disposed below transport path 143 and a light-receiving section (see FIG. 54: light-receiving section 52) disposed above, and detects the basis weight of sheet S based on the intensity of light received by the light-receiving section. Details of basis weight sensor 50 will be described later with reference to FIG. 54.

[0046] (Surface Sensor 60) Next, the configuration of the surface property sensor 60 will be described with reference to Fig. 2, Fig. 3, and Fig. 4 to Fig. 8. Fig. 4 is a cross-sectional view of the surface property sensor 60, and Fig. 5 is a perspective view showing the internal configuration of the surface property sensor 60. Note that in Fig. 5, the cover (housing 61) that covers the entire surface property sensor 60 is omitted.

[0047] As shown in these figures, the surface sensor 60 includes a housing 61, a light emitting unit 62, a collimating lens 63, a plurality of light receiving units 64 (light receiving units 641, 642), and an opening / closing mechanism 65. The housing 61 covers the other components and blocks external light. The housing 61 is not on the bottom surface, and in the attached state, the upper guide 1431 functions as a member that covers the bottom surface of the surface sensor 60. The light emitting unit 62, the collimating lens 63, and the plurality of light receiving units 64 (light receiving units 641, 642) will be described in detail later. In this embodiment, an example in which a collimating lens is used as the optical system disposed between the light source and the irradiation area is shown, but a lens other than a collimating lens may be disposed. For example, when a bullet-shaped LED is used, an optical lens is mounted on the LED. In addition, a convex lens or the like may be disposed between the LED and the collimating lens.

[0048] The opening / closing mechanism 65 includes a shutter 651, a connection portion 652, a rotating shaft 653, a worm gear 654, and a drive motor 655. The power of the drive motor 655 is transmitted to the shutter 651 through the worm gear 654, the rotating shaft 653, and the connection portion 652. The shutter 651 moves around the rotating shaft 653 in the direction of the arrow. The upper guide 1431 has an opening a1, and the lower guide 1432 has an opening a20. FIGS. 4 and 5 show a normal state in which the paper characteristics of the paper S are not measured, and in this case, the opening a1 is closed by the plate-like shutter 651, which is a flat plate member. The opening and closing operation of the opening a1 of the shutter 651 is performed by the control portion 11 controlling the drive motor 655.

[0049] FIG. 6 is a schematic cross-sectional view of the surface property sensor 60 with the shutter open. FIG. 6 shows a state of the measurement mode in which paper characteristics are measured. Note that FIG. 6 omits the configuration of the opening / closing mechanism 65 other than the shutter 651. The opening a1 has a substantially rectangular shape, and the hole size is, for example, several tens of mm in the horizontal direction (X direction) and several tens of mm in the vertical direction (Y' direction). In order not to interfere with the transport of the paper S on the transport path 143, a plurality of ribs r1 (see FIG. 6) are provided on the lower surface of the shutter 651. Each rib r1 is inclined along the transport direction so as to recede on the upstream side of the transport direction and protrude slightly on the downstream side with respect to the lower surface of the upper guide 1431. Such ribs r1 allow the paper S to be transported smoothly. In addition, a reference plate 6501 is attached to the upper surface of the shutter 651 as a member for calibration. The reference plate 6501 is, for example, a white plate with a predetermined surface property, and before measuring the paper characteristics, the surface property sensor 60 is calibrated by measuring the reference plate 6501 with the shutter 651 in the closed state.

[0050] The opening a20 is wider than the opening a1 and is located at approximately the same position in the XY' plane, and the opening a20 encompasses the opening a1. The paper pressing unit 70 is disposed below the lower guide 1432. The paper pressing unit 70 is composed of a pressing plate 71 having an upper surface parallel to the XY' plane, and a drive mechanism (not shown) that moves it up and down. The drive mechanism is composed of a drive motor, a cam, a spring, and the like. Normally, the pressing plate 71 has a shape with a cut portion (inclined surface) that is beveled at the corner on the upstream side of the conveying direction so that the conveyed paper S does not get caught on the corner, and the upper surface of the pressing plate 71 (excluding the cut portion) is located at the same height as the paper passing surface of the lower guide 1432 or a height slightly lower than this. The upper surface of the pressing plate 71 is sufficiently larger than the opening a1. In the measurement mode of FIG. 6, the leading edge of the paper S is conveyed in the conveying direction to a position beyond the openings a1 and a20, and then the paper S is temporarily stopped. After that, the pressing plate 71 is lifted by the drive mechanism of the paper pressing unit 70. As a result, the paper S is fixed with a predetermined biasing force between the upper surface of the pressing plate 71 and the lower surface (reference surface) of the upper guide 1431. For example, the upper surface of the pressing plate 71 is larger than the opening a1 by a width of more than ten millimeters all around, and presses the paper S in the surrounding width region.

[0051] FIG. 7 is a schematic diagram showing the arrangement of the light-emitting unit 62 and the multiple light-receiving units 64, and FIG. 8 is a schematic cross-sectional view showing the angle of incidence (irradiation angle) of the light irradiated by the light-emitting unit 62 and the arrangement angle of the light-receiving units. In this embodiment, the arrangement angle of the light-emitting unit 62 is set so that the angle of incidence of the irradiated light with respect to the reference plane is 75°. This angle of incidence of 75° is an angle used in measuring the glossiness of white paper according to JIS, and is an angle that is less affected by the color of the object to be measured. As described above, the reference plane is a virtual plane including the lower surface of the upper guide 1431, and during measurement, the surface of the paper S, which is the object to be measured, is placed on the reference plane. The light-emitting unit 62 is placed on the substrate b1. The light-emitting unit 62 includes a light-emitting element as a light source, such as an LED that emits light of a predetermined wavelength (for example, an average wavelength of 445 nm or more and 500 nm or less), and the irradiation light emitted from the light source (point light source) is made into approximately parallel light by the collimating lens 63 and is irradiated to the irradiation area. The irradiation area is the inner area of ​​the aperture a1 when viewed from the Z' direction, and the center (optical axis) of the irradiation area intersects with a reference plane parallel to the XY' plane at an intersection p1. A surface-emitting LED or a bullet-shaped LED may be used as the light-emitting unit 62. When a bullet-shaped LED is used, a desired irradiation diameter (also called beam diameter) can be obtained by designing a lens that matches the directivity of the bullet type.

[0052] Each of the multiple light receiving sections 64 includes a light receiving element such as a photodiode or a phototransistor, and includes a first light receiving section 64 (light receiving section 641) that receives specularly reflected light from the irradiation area, and one or more second light receiving sections 64 (light receiving sections 642) that receive diffusely reflected light from the irradiation area. As shown in FIG. 7 (see also FIG. 8 for angles of incidence and reflection), the first light receiving section 641 is disposed at a position of a reflection angle of 75° corresponding to the incident angle of 75° of the light emitting section 62, and receives specularly reflected light. The second light receiving section 642 can be disposed at any reflection angle position except for the position of 75° within a range of reflection angles of 0° or more and less than 90°, and receives diffusely reflected light. The preferred positions are reflection angles of 60°, 30°, and 0°, and more preferably two positions of 60° and 30°, or one position of 60°. 4, 6, and 7 show examples in which a first light receiving unit 641 for receiving specularly reflected light with a reflection angle of 75° and a second light receiving unit 642 for receiving diffusely reflected light with a reflection angle of 30° are arranged. In these figures, the light receiving unit 641 is arranged on the substrate b2, and the light receiving unit 642 is arranged on the substrate b3.

[0053] The housing 61 is provided with openings a3 and a4 on the light receiving paths of the light receiving units 641 and 642. The openings a3 and a4 have the same structure, so the structure of the opening a3 will be described below as a representative example. As shown in the partially enlarged view of FIG. 7, the opening a3 is a circular slit with a diameter of 3 mm when viewed from the intersection point p1 side, and an inclined surface 61a is provided on the light incident side. The inclined surface is configured so that the distance from the optical axis increases toward the incident side, and the overall shape is a mortar shape. The inclination angle is about 45° with respect to the incident angle of the light. By providing such an inclined surface 61a, it is possible to prevent the light reflected at the edge of the opening a3 (slit) from entering the light receiving unit 641, and thus it is possible to accurately detect the characteristics of the recording medium.

[0054] (Irradiation diameter) Fig. 9 is a schematic diagram showing the variation in the microscopic fiber orientation of paper, and Fig. 10 is a schematic diagram showing the state of pulp fibers on the surface of paper. Generally, paper is made by randomly intertwining pulp fibers extracted from plants such as wood in all directions (isotropically), which results in a non-uniform surface with uneven distribution of surface properties (also known as "formation" in industry terms). This formation varies depending on the length and thickness of the pulp fibers in the paper. Compared to Fig. 9A, Fig. 9B shows greater unevenness. Pulp fibers derived from coniferous and broad-leaved trees are used as the material for paper, and the pulp fibers have an average length of 3.32mm to 0.79mm (maximum of about 5.7mm) and an average width (thickness) of 39μm to 19μm (maximum of about 97μm) (reference: paper "Measurement of the Surface Shape of Paper" (Paper and Pulp Technology Association Journal, Vol. 18, No. 2, February 1964), Oji Paper Co., Ltd. (Central Research Institute, Yukinori Hata) In addition, as shown in Figure 10, the surface becomes uneven due to the length of the pulp fibers and the occurrence of bundling fibers (fiber entanglements) between the pulp fibers.

[0055] Therefore, if the irradiation diameter is narrowed too much compared to the length of the pulp fibers or the shape of the bundled fibers, the position of the reflected light on the paper, i.e., the unevenness of the paper surface, will cause differences in the amount of detected specular reflected light and diffused light, resulting in a sensor that is highly sensitive to the paper surface condition even on the same paper surface. This will cause errors in the amount of reflected light, and if the data containing these errors is used to determine the paper type, there is a risk of misidentifying the paper type. For example, matte coated paper will be misidentified as fine paper.

[0056] FIG. 11 is a schematic diagram showing the surface distribution of the paper and the diffusion state of the irradiated light, and FIGS. 12(a) and 12(b) are schematic diagrams showing the irradiation diameter of the irradiated light and the arrangement position of the light emitting unit 62, with FIG. 12(a) being a side view and FIG. 12(b) being a top view. In this embodiment, in view of such a situation, irradiating a wide area as shown in FIG. 11(b) can reduce the influence of the distribution of the paper surface properties, rather than irradiating a narrow area as shown in FIG. 11(a). In other words, by setting the irradiation diameter of the irradiated light to a size that irradiates an area sufficiently wider than the area where the pulp fibers that cause the distribution of the paper surface properties are intertwined, the amount of reflected light of the paper surface properties including the unevenness of the paper surface can be averaged, and the influence of the distribution of the paper surface properties can be reduced. Specifically, the irradiation diameter in the irradiation area is set to 6 mm or more, which is larger than the maximum length of the pulp fibers described above, 5.7 mm. The irradiation diameter here is the diameter of the light (beam) on a plane perpendicular to the optical axis at the intersection p1 where the optical axis intersects with the irradiation surface (reference surface). In general, the focal length of the lens, the size of the light source, and the spread of the irradiation diameter are (Irradiation diameter) ≒ (Light source size) / (Lens focal length) (1) From this formula, the longer the focal length, the closer the light can be to parallel light, but the amount of light emitted decreases, so the lens must be brighter. The general formula for lens brightness is: (Lens brightness) = (Lens focal length) / (Lens diameter) (2) From equation (1), it is possible to widen the irradiation diameter by having a certain size of the light source. In other words, in order to set the irradiation diameter and irradiation light amount, the above-mentioned conditions are obtained from the optical design.

[0057] In this embodiment, the irradiation diameter is set to φ13 mm as shown in FIG. 12(a). In order to achieve such an irradiation diameter, when lens specifications are set to a focal length of 20.7 mm and a lens diameter of φ7.5 mm based on optical design and optical simulation, the distance from the light emitting unit 62 to the intersection point p1 is in the range of 50 to 70 mm, for example, 60 mm. When the irradiation diameter is designed in this way, as shown in FIG. 12(b), when viewed from above (on the XY' plane), the short diameter (length in the Y' direction) is φ13 mm and the long diameter (length in the X direction) is 50.2 mm (=13 / cos75°). With such an irradiation area (512.6 mm 2 ) makes it possible to detect the amount of specularly reflected light or diffused light with low sensitivity to the distribution of the paper's surface properties, thereby reducing the effect of non-uniform fiber orientation.

[0058] FIG. 13 is a graph showing the variation in the amount of detected light depending on the irradiation diameter when measuring multiple points on the paper surface. The figure shows the variation in the amount of detected light when measuring 10 different positions on a single sheet of plain paper, from samples No. 1 to No. 10, with two irradiation diameters, φ13 mm and φ1 mm. The ±8% shown in the figure is the threshold at which false detection of paper type is likely to occur, as explained below. If the irradiation diameter is small, it is more susceptible to the influence of the texture of the paper S, and false detection of paper type is more likely to occur. As shown in FIG. 13, the variation in false detection is smaller with an irradiation diameter of φ13 mm than with an irradiation diameter of φ1 mm, and is within the ±8% threshold.

[0059] FIG. 14 is a graph showing the transition of detection accuracy when errors are added to input data when paper type discrimination is performed using a machine learning model (FIGS. 18A to 20 described later). In the figure, an example is shown using "coated ball" gloss coated paper as the paper type. "Coated ball" was used because it is the paper that is most likely to cause misjudgment at a reflection angle of 75° among 58 types of paper including "coated ball". The graph shows the threshold value (negative side of the X-axis error) at which the first candidate for judgment is misjudged as a different matte coated paper from the original gloss coated paper (correct judgment) when an error with an absolute value larger than -8% is input, and therefore a misjudgment occurs for the first candidate for judgment, so the threshold value of the specular reflection light detection unit for a reflection angle of 75° is set to ±8%. In other words, the sensor detection accuracy for a reflection angle of 75° is set to within ±8%. The above has been explained about a reflection angle of 75°, but similar verification was also performed for reflection angles of 30° and 60° (not shown), and as a result, the threshold at which the first candidate would be misjudged was greater than ±8%, and based on the verification results, the threshold for each of the reflection angles of 30° and 60° was adopted as ±15%. In other words, the sensor detection accuracy for reflection angles of 30° and 60° was set to within ±15%.

[0060] (Light source position relative to focal length) Referring again to FIG. 12(b), the position of the light emitting unit 62 as the light source of the surface sensor 60 (light sensor device) according to this embodiment will be described. A collimating lens 63 is disposed immediately downstream of the light emitting unit 62. In general, a point light source is disposed at the focal position of the collimating lens, that is, the light source is disposed at the same distance as the focal length from the collimating lens on the optical axis, so that the light emitted from the point light source is made into parallel light. In Example 1, the light source is disposed at the focal length position in this way. However, in Example 2, the light emitting unit 62 is disposed closer to the collimating lens 63 at a distance shorter than the focal length. That is, the focal length f is set to be greater than the light source distance t0 (here, t0 is the distance between the light emitting unit 62 and the center position of the collimating lens 63 on the optical axis, and f is the focal length of the collimating lens 63).

[0061] 15A and 15B are graphs showing the robustness of the mounting accuracy of the light emitting unit 62. FIG. 15A is a diagram showing the output error sensitivity to the mounting error in the rotation direction, and FIG. 15B is a diagram showing the output error sensitivity to the mounting error in the optical axis direction. Data is also shown for Example 1, which is a comparative example, in addition to Example 2, in which the light source is arranged at the focal length. Here, the mounting error in the rotation direction is the angle when rotating around an axis perpendicular to the optical axis (corresponding to YAW). The mounting error in the optical axis direction is the distance in the front-rear direction along the optical axis. As shown in FIGS. 15A and 15B, it can be seen that Example 2, in which the light source is arranged at a distance shorter than the focal length of the collimating lens, has less fluctuation with respect to errors, is more robust, and is superior to Example 1.

[0062] (wavelength of light from the light source) The wavelength of light from the light emitting unit 62 is selected as a light source with a blue wavelength that is not absorbed by the pigment in the surface layer of the coated paper. From the experimental verification of the LED, the following results were obtained. 405nm (purple): Coated paper can be distinguished (specular reflection). With diffuse light, it is difficult to distinguish between high quality and normal paper. 465nm (blue): Coated paper can be distinguished (specular reflection). With diffuse light, it can be distinguished at 30° and 60°. 525nm (green): Coated paper can be distinguished (specular reflection). With diffuse light, it can be distinguished at 60°. 680nm (red): Difficult to distinguish, especially in diffuse light.

[0063] In addition, the correlation with each wavelength was also examined with respect to the correlation with Beck smoothness as shown below.

[0064] Fig. 16 is a table showing the degree of correlation between the wavelength of light from the light emitting unit 62 and Beck smoothness, and Fig. 17 is a graph showing the comparison results between the Beck smoothness of 12 types of paper (gloss coated paper, matte coated paper, plain paper, and fine paper) and the predicted value based on the measurement results. Here, Beck smoothness (sec) is a measurement method specified by JIS, in which a pressure of 0.1 MPa is applied from above, the pressure is reduced to half atmospheric pressure using a vacuum pump, and the time it takes for 10 cc of air to flow through the gap between the measurement sample surface and the glass surface over a distance of 13 mm is measured.

[0065] The light emitting unit 62 was placed at a position with an incident angle of 75°, and three types of LEDs with wavelengths of 405 nm, 465 nm, and 525 nm were used as light sources. The light receiving unit 64 was placed at three locations with reflection angles of 30, 60, and 75° (in the test with a wavelength of 525 nm, it was placed at a total of four locations, plus at 0°).

[0066] Four types of paper were used: gloss coated paper, matte coated paper, plain paper, and fine paper, with four different brands of each (a total of 16 brands of paper). On the horizontal axis of Figure 17, the four brands of gloss coated paper are indicated as G1 to G4, the four brands of matte coated paper as M1 to M4, the four brands of plain paper as R1 to R4, and the four brands of fine paper as F1 to F4.

[0067] As shown in Fig. 16, when the correlation between the predicted smoothness value calculated by multiplying the measured value from each light receiving unit 64 by a predetermined coefficient and adding them up for each of 16 brands of paper and the measured value (log) of the Beck smoothness of each paper was compared, it was found that the value of the multiple correlation R was the highest when an LED with an average wavelength of 465 nm was used (shown in a dashed frame). From this, it is preferable that the wavelength of the light source is in the range of more than 405 nm and less than 525 nm, more preferably in the range of 445 nm or more and 500 nm or less, and the most preferable wavelength is around 465 nm.

[0068] Figure 17 shows the relationship between the predicted value and the Beck smoothness (log) when using a 465 nm LED. The predicted value here is Predicted value = 3.4518 x V75 - 3.9595 x V60 + 1.5383 x V30 - 0.4789, where V75, V60, and V30 are the output values ​​of the light receiving unit 64 at the reflection angles of 75, 60, and 30 degrees, respectively. As shown in Figure 17, it can be seen that there is a high correlation.

[0069] In the optical sensor device (surface property sensor 60) according to this embodiment, the irradiation diameter in the irradiation area of ​​the light irradiated by the light emitting unit 62 is set to an irradiation diameter that provides an irradiation area that is less affected by non-uniformity in the fiber orientation of the recording medium. More specifically, the irradiation diameter (minor diameter) in the irradiation area (intersection p1) is set to 6 mm or more, which is longer than the maximum length of the pulp fibers. This results in an optical sensor device that can accurately detect the characteristics of the recording medium without being affected by the surface property distribution of the recording medium.

[0070] In addition, a collimating lens 63 is disposed immediately downstream of the light emitting unit 62, and the light emitting unit 62 is disposed on the optical axis at a distance shorter than the focal length of the collimating lens 63. This makes it possible to increase the robustness of the light emitting unit 62 with respect to the installation accuracy.

[0071] Moreover, by setting the emission wavelength of the light emitting portion 62 to be 445 nm or more and 500 nm or less, the surface property can be detected with high accuracy. Specifically, it is possible to obtain detection data that is highly correlated with the Beck smoothness.

[0072] Furthermore, the optical sensor device according to this embodiment is provided with a shutter 651 that opens and closes the opening a1 that constitutes the irradiation area. For example, if the optical sensor device is disposed on a paper transport path, paper dust from the transported paper S may adhere to optical components such as the collimator lens 63 and the light receiving unit 64, resulting in a decrease in measurement accuracy. When the surface properties of the recording medium are not being measured, closing the opening a1 with the shutter 651 prevents the optical components of the optical sensor device from being soiled by paper dust, and thus prevents a decrease in measurement accuracy.

[0073] (Paper type determination process) Next, the paper type determination process performed by the image forming apparatus 10 will be described with reference to Fig. 18A to Fig. 29. Fig. 18A is a block diagram showing the configuration of the image forming apparatus. Fig. 18B is a block diagram showing the functional configuration of an overall control unit of the image forming apparatus according to the first embodiment. Fig. 19 is a flowchart showing the print process of the image forming apparatus.

[0074] 18A, image forming apparatus 10 is connected to server 80 and other image forming apparatuses 10b and 10c via network L. In the figure, configurations other than control unit 11 of image forming apparatus 10 have already been explained in FIG. 1 and the like, so explanations are omitted by assigning the same reference numerals.

[0075] The control unit 11 functions as an overall control unit 110, an engine control unit 120, a media sensor control unit 130, a post-processing option control unit 140, a paper feed option control unit 150, and a transport control / image formation control unit 160. The date and time management unit 121 manages the date and time used in the image forming apparatus 10. When the control unit 11 executes a process that requires the date and time, the control unit 11 acquires the date and time from the date and time management unit 121 and executes various processes. For example, when determining the execution timing of the generation or update process of a paper type discrimination model or a paper type discrimination algorithm described later, the date and time managed in the date and time management unit 121 may be used. The date and time managed in the date and time management unit 121 may be updated and adjusted as appropriate by communicating with an external time adjustment server (not shown) or the like via a network, for example.

[0076] 18B, the overall control unit 110 functions as an acquisition unit 111 and a discrimination unit 112. The acquisition unit 111 acquires a specular reflection light amount value relating to the amount of specular reflection light specularly reflected from the surface of the paper, a diffuse reflection light amount value relating to the amount of diffuse reflection light diffusely reflected from the surface of the paper, a value relating to the thickness of the paper, and a value relating to the basis weight of the paper, etc. The discrimination unit 112 inputs the specular reflection light amount value, diffuse reflection light amount value, value relating to the thickness of the paper, and value relating to the basis weight of the paper acquired by the acquisition unit 111 into a machine-learned model for paper type discrimination to discriminate the type of paper.

[0077] When a print job is input by an instruction sent from an external terminal such as the operation panel 18 or a network-connected PC operated by a user, the overall control unit 110 causes the engine control unit 120 to execute the print job based on the print setting information of the input print job. The overall control unit 110 executes a paper type determination process using a paper type discrimination engine (a paper type discrimination model that is a learned model) and a paper profile stored in the storage unit 12. Here, the "paper profile" is a pre-registered value that corresponds to the measured values ​​of the paper physical properties obtained from the media sensor 15, the calculated values ​​of the measured values, the characteristic data input by the user, the paper size, and an arbitrary identification name (e.g., paper brand), etc. The measured values ​​and calculated values ​​from the media sensor 15 include, for example, the measured values ​​of the reflected light at 75° and 30° of the surface property sensor 60 of the paper physical properties, the calculated value of the basis weight and / or the calculated value of the basis weight difference based on the measured value of the basis weight sensor 50, and the calculated value of the paper thickness based on the measured value of the paper thickness sensor 40, which will be described later. In addition, the characteristic data input by the user includes various control parameter values ​​such as conveying system characteristic data (front / back adjustment values, offset correction values, etc.), image formation characteristic data (fixing temperature values, density adjustment values, gamma correction values, etc.), paper feed system characteristic values ​​(paper feed tray handling fan air volume adjustment values), etc.

[0078] The "paper type discrimination engine" is also called a trained model, and is a trained model generated by supervised learning using training data with the detection output of the media sensor 15 of the paper S as an input value and the paper type information set by the user of the paper S as a correct answer label. As the training data, data from other image forming apparatuses 10b, 10c, etc. connected to the network L may be aggregated by the server 80. The trained model is generated by a learning method using, for example, a random forest. Note that the learning method is not limited to this, and various methods can be used as long as they are supervised learning. For example, a neural network, a support vector machine (SVM), boosting, a Bayesian network linear discrimination method, a nonlinear discrimination method, etc. can be applied. In addition, a standalone high-performance computer using a CPU and a GPU (Graphics Processing Unit) processor, or a cloud computer can be used as a learning machine that executes machine learning. The method of generating a paper type discrimination model by machine learning will be described in detail later.

[0079] The engine control unit 120 performs processing related to image formation by controlling the post-processing option control unit 140, the paper feed option control unit 150, and the transport control / image formation control unit 160. The post-processing option control unit 140 controls the post-processing device 30. Specifically, it transmits paper transport timing, post-processing setting information for the transported paper, etc. to the post-processing device 30. The paper feed option control unit 150 controls the paper feed device 20. Specifically, it communicates with the paper feed device 20 to transmit and receive information on the paper feed tray to be used, the paper transport timing, etc.

[0080] The transport control / image formation control unit 160 controls the paper feed transport unit 14 (including drive motors for the transport paths 143, 144, the fixing unit 136, etc.) to control the feeding and transport of the paper S. It also controls the image forming unit 13 to control the image formation conditions and the image formation timing according to the paper position.

[0081] The media sensor control unit 130 controls the paper thickness sensor 40, the basis weight sensor 50, and the surface property sensor 60 in response to an execution instruction request from the engine control unit 120, and causes them to measure the paper characteristics. The media sensor control unit 130 also controls the operation of the paper pressing unit 70.

[0082] Next, the paper type determination process will be described with reference to Fig. 19. Steps S10 to S30 are print preparation processes. The user performs this print preparation process before carrying out actual printing of a print job.

[0083] (Step S10) The user operates a paper setting button on an operation screen (not shown) displayed on operation panel 18. Control unit 11 starts paper setting by accepting this operation from the user. This instruction to start paper setting includes selection information for one or more paper feed trays (paper feed trays 141, 142, 241 to 243) in which the target paper is loaded. Details of this paper setting process will be described later with reference to FIG. 20.

[0084] (Step S20) Upon completion of the paper settings, the image forming conditions are set according to the paper characteristics that have been set, and a print button (start button) not shown is pressed to perform a test print (trial print) of the print job.

[0085] (Step S30) If the user is not satisfied with the test printing results, or if multiple types of paper are used in one print job, the user repeats the process from step S10 onwards for another type of paper (step S30: NO).On the other hand, if the user is satisfied with the test printing results and confirmation of all paper types has been completed (YES), control unit 11 accepts the user's operation to complete preparation and advances the process to step S40.

[0086] (Step S40) The control unit 11 controls the image forming unit 13 and the like to execute the print job (main printing), thereby completing the print process (end).

[0087] (Paper setting process) (Steps S100 and S101) 20 is a subroutine flowchart showing the paper setting process (step S10). The transport control / image formation control unit 160 of the control unit 11 causes the opening / closing mechanism 65 to open the shutter 651, and then feeds the paper S from the paper feed tray selected by the user (for example, the paper feed tray 141 (hereinafter simply referred to as the "selected tray")) and transports it to the transport path 143.

[0088] (Step S102a) The media sensor control unit 130 detects the thickness of the paper S using the paper thickness sensor 40, and passes the detected data (hereinafter, referred to as “measured value 1”) to the overall control unit 110.

[0089] (Steps S102b and S102c) The media sensor control unit 130 detects the basis weight and surface properties of the paper using the basis weight sensor 50 and the surface property sensor 60, and passes the detection data (hereinafter referred to as "measured values ​​2 and 3," respectively) to the overall control unit 110. In this measurement, when the leading edge of the paper S passes by a predetermined amount through the detection positions of the basis weight sensor 50 and the surface property sensor 60, the paper transport is temporarily stopped, the pressing plate 71 of the paper pressing unit 70 is lifted, and the paper S located in the irradiation area of ​​the surface property sensor 60 is fixed. Note that before this first measurement, the surface property sensor 60 may be calibrated using a reference plate 6501.

[0090] (Step S103) If the predetermined number N of measurements have not been performed, the processes of steps S102b and S102c are repeated. Note that, each time before the next measurement, the paper pressing unit 70 is moved up and down and the paper S is transported a predetermined distance (for example, an arbitrary distance of about several mm to 50 mm) to measure a different position on the same paper S. N times is, for example, five times, and if five measurements have been completed (YES), the process proceeds to step S104.

[0091] (Step S104) When the rear end of the paper S passes the media sensor 15, specifically, when the paper sensor (not shown) detects that the rear end of the paper S has left the transport roller 1434 (YES), the process proceeds to step S105.

[0092] (Step S105) Here, the control unit 11 closes the shutter 651 using the opening / closing mechanism 65 to keep it in a closed state. This prevents the surface property sensor 60 from being soiled by paper powder or the like caused by the paper S being transported when the media sensor 15 is not performing a measurement.

[0093] (Step S106) Here, the detection data (measurements 2 and 3) of basis weight and surface property for N times acquired in steps S102b and S102c are averaged.

[0094] (Step S107) The overall control unit 110 performs paper type determination and basis weight classification determination using the measurement values ​​1 to 3 (or their average data) acquired in steps S102a to S102c, the trained model (paper type discrimination engine), and basis weight classification probability calculation processing. Also, it performs candidate paper profile determination with high similarity from registered paper profile data using the measurement values ​​1 to 3 (or their average data) acquired in steps S102a to S102c and profile selection processing.

[0095] The paper type and basis weight classification determination processes will be described in more detail with reference to Fig. 21. Fig. 21 is a control block diagram showing the determination process (S107). In addition to the paper type and basis weight classification determination processes, Fig. 21 also shows the determination process of registered profile candidates determined using a paper profile created for each user (each device).

[0096] (S701~S703) The overall control unit 110 obtains a basis weight conversion value, a paper thickness conversion value, and a surface property measurement value using the measurement values ​​1 to 3 acquired in steps S102a to S102c. Here, the basis weight conversion value is calculated from the first basis weight and the second basis weight using a coefficient and a formula determined by the surface property measurement value of the measurement value 3 (S703) and the ambient environment information (temperature, humidity) of the image forming system 1. Here, the basis weight difference = first basis weight - second basis weight. The basis weight sensor 50 is equipped with a plurality of LEDs that irradiate irradiation light of different wavelengths. The first basis weight is calculated by using a first LED that outputs irradiation light of a wavelength (750 nm to 900 nm) and the amount of transmitted light that passes through the paper S of this irradiation light. The second basis weight is calculated by using a second LED that outputs irradiation light of a wavelength (400 nm to 470 nm) and the amount of transmitted light that passes through the paper S of this irradiation light. The basis weight is sent to steps S711, S712, and S714, and the basis weight difference (first basis weight - second basis weight) is sent to steps S713 and S714. The paper thickness is sent to steps S712 and S713. The measured surface properties are sent to steps S701, S713, and S714.

[0097] (Step S711) The overall control unit 110 calculates the basis weight category probability from the basis weight calculated in step S701. Examples of basis weight categories are the following 12 categories. ~61g / m 2 62~75 g / m 2 76~81 g / m 2 82~92 g / m 2 93~106 g / m 2 107~136 g / m 2 137~177 g / m 2 178~217 g / m 2 218~257 g / m 2 258~301 g / m 2 302~351 g / m 2 352 g / m 2 The calculated basis weight is assumed to follow a normal distribution and vary with a predetermined standard deviation, and the probability of each division is determined. For example, if it is close to the center of one of the divisions, the division probability is high and close to 100%. On the other hand, the further away from the center of the division, i.e., the closer to the boundary, the lower the probability. The division probability is sent to step S721 as the basis weight division score.

[0098] (Step S712) The overall control unit 110 uses the basis weight and the paper thickness to calculate the density (=basis weight / paper thickness). The calculated density is sent to step S713.

[0099] (Step S713) The overall control unit 110 performs paper type discrimination using the paper characteristic data of the measured values ​​of density, paper thickness, and surface properties, and the trained model. The discrimination result is sent to step S721 as a paper type score.

[0100] (Step S714) The overall control unit 110 uses the paper characteristic data of density, paper thickness, and surface property measurements, and a list of paper profiles registered in advance by the user, etc., to select a registered profile with a high compatibility rate from among them. This list of paper profiles has data representing paper physical properties consisting of basis weight value, paper thickness value, basis weight difference value, and surface property measurement value (for example, measurement values ​​from the light receiving units 641 and 642), and the overall control unit 110 selects a paper profile candidate from the registered paper profiles in order of closestness to the registered paper physical property data using the measurement values ​​1 to 3 (or their calculation data) acquired in steps S102a to S102c. As a method of determining the closeness, for example, the overall control unit 110 calculates the distance between the values ​​of each data item, and determines that the closest one is closer in order of the sum of the values ​​obtained by multiplying the calculation result by a weighting coefficient, and selects the closest one as a candidate. The selection result is sent to step S722 as a candidate paper profile list to which a compatibility score is assigned.

[0101] (Step S721) The overall control unit 110 displays one or more paper type / basis weight candidates in descending order of probability according to the basis weight classification score and the paper type score, and presents them to the user. Details of this process will be described later with reference to FIG. 25.

[0102] (Step S722) Depending on the matching score, overall control unit 110 displays one or more registered profile candidates in descending order of score and presents them to the user. Details of this process will be described later with reference to FIG.

[0103] (Step S108) Referring again to Fig. 20, general control unit 110 displays the determination result. This process corresponds to Fig. 22, Fig. 24, and steps S721 and S722 described above.

[0104] (Step S109) If the user wishes to change the paper type (YES), the process proceeds to step S110; if the user wishes to accept the determination result without making any changes (NO), the process proceeds to step S111.

[0105] (Step S110) The overall control unit 110 accepts an input operation from the user via the operation panel 18, and sets the accepted changed paper type as the paper type information of the selected tray.

[0106] (Step S111) If there is no change in the settings (for example, when the operation of the button 84 in FIG. 22 is accepted), the overall control unit 110 sets the result of the determination made in step S107 as the paper type information of the selected tray.

[0107] (Step S112) The printing conditions for the selected tray are set to the printing conditions corresponding to the set paper type, and then the process returns to the process in FIG. 19, and the processes from step S10 onwards are carried out (return).

[0108] Next, the procedure of the process executed by the user in the above-mentioned paper setting process will be described, including examples of screens displayed on the image forming apparatus 10.

[0109] FIG. 22 is a diagram for explaining the procedure of the paper setting process. FIG. 23 is an example of an operation screen for receiving an instruction to start the determination process. FIG. 24 is an example of an operation screen for receiving an instruction to execute the determination process. FIG. 25 is an example of an operation screen showing the determination result (paper type / basis weight classification). FIG. 26 is an example of an operation screen showing the determination result (registered profile). FIG. 27 is an example of an operation screen for receiving an instruction to start a change process of the determined paper settings. FIG. 28 is an example of an operation screen for receiving an instruction to change the determined paper settings. FIG. 29 is an example of an operation screen for receiving an instruction to register the change of the determined paper settings.

[0110] (Step S1) As shown in FIG. 22, first, the user places the paper to be measured in the tray, and selects the automatic paper setting button 180a on a screen such as that shown in FIG. 23 displayed on the operation panel 18 of the image forming apparatus 10.

[0111] (Step S2) A screen such as that shown in FIG. 24 is displayed on the operation panel 18. When the user selects the tray in which the paper is set and selects the paper type detection button 180b, the paper setting process is started.

[0112] (Step S3) Next, the image forming apparatus 10 executes the paper conveyance, measurement, and judgment processes shown in FIG.

[0113] (Step S4) Next, the judgment result is displayed on the screens shown in FIGS.

[0114] (Step S5) The user selects the one to be used from the displayed candidates and applies the settings to the image forming apparatus 10.

[0115] FIG. 25 is an example of an operation screen showing the judgment result (paper type / basis weight classification) displayed on the operation panel 18. In the operation screen of FIG. 25, two candidates of paper type / basis weight are displayed in descending order of score. When the user accepts the judgment result of the first candidate, column 181 (gloss coated paper / basis weight 177 to 216), the user operates button 184, and the judgment result of the selected column 181 is applied to tray 1 (paper feed tray 141). When the user applies the second candidate, column 182 (plain paper / basis weight 172 to 216), the user selects column 182 and then operates button 184. On the other hand, when the user wants to apply the selected judgment result to both trays 1 and 2, the user operates button 185. Also, by operating button 183, the judgment result can be discarded without being adopted. When the user performs a test print and is not satisfied with the image quality or transport, the user reselects the second candidate, column 182, and operates button 184 or button 185 to reapply it to the tray before performing a test print. The judgment result selected by the user may be stored as correct answer data in the storage unit of the image forming apparatus 10 or the server 80, etc., and may be used as training data for subsequent machine learning. In addition, corrections may be accepted by manual settings in which the user directly inputs various settings.

[0116] FIG. 26 is an example of an operation screen showing the judgment result (registered profile) displayed on the operation panel 18. The operation screen of FIG. 26 displays three candidates of registered profiles in descending order of score. The registered profile is registered by associating the paper characteristics previously measured by the media sensor 15 or the paper characteristics manually input through the operation panel 18 with the media name. If the paper characteristics are the same as those previously measured, the registered media name is displayed as the first candidate. When the judgment result of the first candidate field 186 (paper profile data 21) is accepted, the judgment result of the selected field 186 is applied to the tray 1, etc., by operating buttons 184, 185, etc. Note that, as candidates of registered profiles, preset profiles registered in advance in the device may be displayed based on the user's instructions, etc. Furthermore, if the image quality or paper feed and transport process does not satisfy the user's requirements as a result of performing test printing (trial printing) in S20 of FIG. 19 using various settings applied based on the judgment result as described above, the user can select the second candidate again and perform test printing. The user may also manually set various setting values. The manually set various settings may be stored in the image forming device 10 or the like, or may be added to a teacher database (to be described later) together with various measured values, according to instructions from the user. The process of determining the registered profile will be described in detail later.

[0117] Here, the user can modify the determined paper setting contents and set them in the image forming apparatus 10. For example, in a paper setting screen as shown in FIG. 27, it is displayed that the paper type has been determined to be "glossy coated paper". When the user selects the setting change button 187 in the screen of FIG. 27, a screen for changing the paper settings as shown in FIG. 28 is displayed. In the screen of FIG. 28, a list of setting items and current setting contents are displayed on the left side. Then, since the paper type item is selected as the item to be changed, the current setting of the paper type, "glossy coated paper", and candidates for changing the paper type are displayed on the right side of the screen of FIG. 28. Here, as in the screen shown in FIG. 29, when "fine paper" is selected from the candidates for changing the paper type on the right side of the screen, the list of setting items on the left side of the screen also displays that the paper type will be changed to "fine paper". In this way, when the user changes the selected paper type (paper type) from "glossy coated paper" to "fine paper" and presses the "OK" or "Register & OK" button, the setting of "fine paper" is applied and registered in the image forming apparatus 10. As described below, the registered content may be added as teacher data for machine learning. Also, a button or the like may be provided separately to receive an instruction from a user to add the content as teacher data for machine learning. As described below, if the paper type is changed to a specific type such as "color paper," "envelope," or "embossed paper," the content may not be added to the teacher data for machine learning.

[0118] In this way, in the image forming apparatus 10 according to the present embodiment, the measured values ​​including the acquired specular reflection light amount value and diffuse reflection light amount value are input to a machine-learned trained model for paper type discrimination to discriminate the paper type. This makes it possible to appropriately discriminate candidate paper types even for paper types that are not registered in advance in the paper type database.

[0119] (Creating a paper type discrimination model using machine learning) Fig. 30 is a diagram showing an example of information registered in the training data database, and Fig. 31 is a diagram showing an example of a trained model of a random forest generated by machine learning using training data.

[0120] As shown in FIG. 30, the input factors of the teacher data used for machine learning are the light amounts obtained from the surface property sensor 60, that is, reflected light 75° (specular reflected light), reflected light 60° (diffuse reflected light), and reflected light 30° (diffuse reflected light), as well as the basis weight difference (first basis weight-second basis weight) and density information obtained from the paper thickness sensor 40 and basis weight sensor 50. Information on basis weight and paper thickness may be used instead of density. Also, paper type (paper type) is used as an output factor of the teacher data. In the example of FIG. 30, more than 800 sets of teacher data are prepared.

[0121] By performing machine learning using a random forest model using training data such as that shown in FIG. 30, a trained model (paper type discrimination model) as exemplified in FIG. 31 is generated. In this embodiment, a paper type discrimination model with about 500 nodes is generated. Note that the generation of a random forest model using training data and a discrimination method using a random forest model are publicly known, so detailed explanations will be omitted. Since the random forest model has a tree-like structure and branches (determines) at each node according to a threshold condition, it is implemented as a simple conditional branch when incorporated into a program as a discrimination algorithm. Therefore, there is an advantage that the processing time required for discrimination is short, and there is no problem even when incorporated as real-time processing in the image forming apparatus 10 or the like. In this embodiment, the generated paper type discrimination model is converted into a paper type discrimination algorithm and incorporated into the program in advance, and executed by the overall control unit 110 of the image forming apparatus 10.

[0122] (Selection of input factors for training data) The present inventors have thoroughly investigated parameters to be used in machine learning to generate a paper type discrimination model that can accurately discriminate the paper type based on information measurable from paper, and have found that the input factors of the training data shown in Fig. 30 are effective. A method for selecting the input factors will be described in detail below.

[0123] First, the accuracy of various physical properties measurable from paper as input factors for the paper type discrimination model was examined. From the perspective of discrimination accuracy, it was found that the contributions were greatest in the order of air permeability, density, Beck smoothness, friction coefficient, stiffness, whiteness, surface resistance, and moisture content (Table 1 below). Furthermore, after confirming that there were no problems with discrimination accuracy even when air permeability, which had the highest contribution, was removed from the input factors, further selection was carried out. Air permeability was excluded from the input factors because an air permeability sensor that can be installed and used in image forming devices and the like is not yet at a practical level.

[0124] [Table 1]

[0125] In the selection, the paper type was actually determined based on the output of each sensor mounted on the image forming apparatus 10, and the effective input factor with the best value in terms of contribution, determination accuracy, and reproducibility was selected. In addition, multiple factors with similar meanings were narrowed down to one because they could cause overjudgment. Density was selected as an input factor because it has a high contribution and can be calculated from the basis weight and paper thickness. As described above, it was found through verification that Beck smoothness is correlated with the amount of specular reflected light and diffuse reflected light measured by the surface sensor 60, so the amount of specular reflected light and diffuse reflected light was selected as an input factor. Data on the ratio of the amount of specular reflected light to the amount of diffuse reflected light was also considered as an input factor, but was excluded as a result of the consideration.

[0126] In this way, the eight input factors were narrowed down to four input factors, and finally, the amount of specular reflection light (75°), the amount of diffuse reflection light (30°), the basis weight difference (first basis weight - second basis weight: amount of light transmitted through the paper) obtained from the basis weight sensor 50, and density (the result calculated from the basis weight and thickness may be used) were selected (Table 2 below). Note that the basis weight difference (first basis weight - second basis weight) was adopted as an input factor because it was found as a result of verification that it can be used to distinguish between gloss paper and non-gloss paper, as described later. The method of obtaining the first basis weight and the second basis weight and an example of using them to distinguish paper types will be described in detail later with reference to Figures 54 to 59. The judgment accuracy of the paper type discrimination model obtained by machine learning using the teacher data created by associating the input factors selected as above with the output factor paper type was 98% (reproducibility 97%) (Table 3 below).

[0127] [Table 2]

[0128] [Table 3]

[0129] Since there is a correlation between the amount of specular reflected light and diffuse reflected light and the smoothness, the smoothness may be calculated by inputting the amount of specular reflected light and diffuse reflected light into the above-mentioned smoothness estimation formula (obtained as a result of composite multiple regression analysis), and the calculated smoothness value may be used as an input factor for machine learning. The values ​​used as the parameters of the training data are not limited to the above examples, and any parameters effective for discriminating paper type may be adopted.

[0130] Second embodiment In the first embodiment, an example is described in which a paper type discrimination process is performed by previously incorporating a paper type discrimination algorithm in the image forming apparatus 10. In the second embodiment, an example is described in which machine learning is performed in the image forming apparatus 10 to generate and update a paper type discrimination model and a paper type discrimination algorithm.

[0131] The configuration of the image forming apparatus 10 of the second embodiment is similar to that of the image forming apparatus 10 of the first embodiment, but differs from that of the image forming apparatus 10 of the first embodiment in that a database of teacher data for performing machine learning as shown in Fig. 30 is stored in the storage unit 12. In this embodiment, the image forming apparatus 10 or the control unit 11 (overall control unit 110) functions as a discrimination device or a discrimination system.

[0132] FIG. 32 is a block diagram showing the functional configuration of an overall control unit of an image forming apparatus according to the second embodiment.

[0133] As shown in FIG. 32, the overall control unit 110 of the image forming apparatus 10 of the second embodiment functions as an acquisition unit 111, a discrimination unit 112, and a learning unit 113.

[0134] The functions of the acquisition unit 111 and the determination unit 112 are similar to those in the first embodiment, and therefore a description thereof will be omitted.

[0135] The learning unit 113 performs machine learning using training data such as that shown in Fig. 30, and generates a paper type discrimination model by a random forest model such as that shown in Fig. 31. Furthermore, when the result of discrimination using the paper type discrimination model (or the converted paper type discrimination algorithm) is changed by the user, the learning unit 113 adds the changed content as training data, and further performs machine learning using the added training data, thereby generating a new paper type discrimination model. The above-mentioned processing of the learning unit 113 will be described in detail below.

[0136] (Adding change information to teacher data) First, the process of adding change information to teacher data will be described.

[0137] 33 is a flowchart showing a process of adding modification information to teacher data, which is executed in the image forming apparatus according to the second embodiment. The following will be described with reference to FIG.

[0138] (Step S201) The overall control unit 110 of the image forming apparatus 10 judges whether or not the result of the judgment by the judgment unit 112 regarding the paper type, etc. has been changed by the user.

[0139] If no change has been made (step S201: NO), overall control unit 110 ends the process.

[0140] If it has been changed (step S201: YES), overall control unit 110 proceeds to the process of step S202.

[0141] (Step S202) The overall control unit 110 sets the changed paper type etc. as the paper setting to be used for image formation.

[0142] (Step S203) The overall control unit 110 acquires corresponding measurement value information. The corresponding measurement value information is information on measurement values ​​such as paper thickness, basis weight (density), and paper surface properties that are input to the trained model in order to determine the paper type, etc. in the discrimination unit 112.

[0143] (Steps S204 and S205) The overall control unit 110 reads the address information of the teacher database stored in the storage unit 12, and acquires the final address of the teacher database.

[0144] (Step S206) The overall control unit 110 adds the paper type information changed by the user and the corresponding measurement value information as new teacher data to the address next to the last address of the teacher database. For example, in the teacher database of Fig. 30, if the teacher data recorded in the last address is teacher data "No. 802", the new teacher data is added as data No. 803. Next, the overall control unit 110 counts up (increments) the update count N1, which is the number of times teacher data has been added, by one, stores it in the memory unit 12, and ends the process.

[0145] (Creating and updating paper type discrimination models and paper type discrimination algorithms) Next, a process for generating and updating a paper type discrimination model and a paper type discrimination algorithm will be described.

[0146] 34 is a flowchart showing a process of generating and updating a paper type discrimination model and a paper type discrimination algorithm executed in an image forming apparatus according to the second embodiment. The following will be described with reference to FIG.

[0147] (Step S211) The overall control unit 110 of the image forming apparatus 10 reads out from the storage unit 12 the number of updates N1 of the teacher data.

[0148] (Step S212).

[0149] The overall control unit 110 determines whether the number of updates N1 of the teacher data is equal to or greater than a predetermined number of updates set in advance in the memory unit 12 as an execution condition for updating the paper type discrimination model. Details of the setting of the execution condition for updating the paper type discrimination model will be described later with reference to Figures 36A and 36B.

[0150] If the number of times N1 the teacher data has been updated is not equal to or greater than the predetermined number of times (step S212: NO), the overall control unit 110 ends the process.

[0151] If the number of times N1 the teacher data has been updated is equal to or greater than the predetermined number of times (step S212: YES), the overall control unit 110 proceeds to the process of step S213.

[0152] (Step S213) The overall control unit 110 sets the number of times N1 the teacher data has been updated to zero.

[0153] (Step S214) The overall control unit 110 judges whether the update timing previously set in the storage unit 12 as an execution condition is a "standby state".

[0154] If the update timing preset as the execution condition is the "standby state" (step S214: YES), general control unit 110 proceeds to the process of step S215.

[0155] If the update timing preset as the execution condition is not "standby state" (step S214: NO), general control unit 110 proceeds to the process of step S216.

[0156] (Step S215) General control unit 110 determines whether image forming apparatus 10 is in a standby state.

[0157] If the image forming apparatus 10 is not in a standby state (step S215: NO), the overall control unit 110 continues the process of monitoring the state of the image forming apparatus 10 until the image forming apparatus 10 is in a standby state. Alternatively, the overall control unit 110 may end the process once and execute the process of FIG. 34 again at a predetermined timing.

[0158] If it is in the standby state (step S215: YES), general control unit 110 proceeds to the process of step S218.

[0159] (Step S216) General control unit 110 judges whether or not the update timing preset as an execution condition is a "designated time".

[0160] If the update timing preset as the execution condition is the "designated time" (step S216: YES), general control unit 110 proceeds to the process of step S217.

[0161] If the update timing preset as the execution condition is not the "designated time" (step S216: NO), the overall control unit 110 ends the process assuming that the update execution condition is not set. Alternatively, the overall control unit 110 may determine the update execution condition based on a predetermined default setting or the like.

[0162] (Step S217) General control section 110 refers to the date and time managed in date and time management section 121, and determines whether or not a preset designated time has arrived.

[0163] If the specified time has not arrived (step S217: NO), the overall control unit 110 continues the process of monitoring the time when the specified time arrives. Alternatively, the overall control unit 110 may end the process once and execute the process of FIG. 34 again at a predetermined timing.

[0164] If the designated time has arrived (step S217: YES), overall control unit 110 proceeds to the process of step S218.

[0165] (Step S218) The overall control unit 110 prohibits the reception of instructions from the user or the like regarding the printing operation or adjustment operation of the image forming apparatus 10.

[0166] (Step S219) The overall control unit 110 generates a paper type discrimination model (FIG. 31) using a random forest based on the teacher database (FIG. 30) stored in the storage unit 12.

[0167] (Step S220) Based on the generated paper type discrimination model, the overall control unit 110 generates a paper type discrimination algorithm that can be executed as a program. The processes in steps S219 and S220 will be described in detail later with reference to FIG.

[0168] (Step S221) The overall control unit 110 writes the generated paper type discrimination algorithm into a designated segment of the flash memory of the CPU that the overall control unit 110 has.

[0169] (Step S222) General control unit 110 permits (cancels prohibition) reception of instructions from a user or the like regarding the printing operation or adjustment operation of image forming apparatus 10 that was prohibited in step S218.

[0170] The overall control unit 110 may manage version information of the generated paper type discrimination model or paper type discrimination algorithm, and display the version information being used on the operation panel 18. The overall control unit 110 may also display a list of the above version information on the operation panel 18 as an update history. For example, a consecutive number is automatically assigned as the version information. The list displays the version information and the update date and time (creation date and time), etc. For example, if the user is not satisfied with the discrimination result by the updated version of the paper type discrimination model, the user can revert to the previous version of the paper type discrimination model by selecting the previous version information displayed in the list. Also, as will be described later with reference to FIG. 36B, the user can revert to the previous version of the paper type discrimination model, or the next version of the paper type discrimination model.

[0171] (Generation of a paper type discrimination model and conversion to a paper type discrimination algorithm) FIG. 35 is a diagram showing a procedure for generating a paper type discrimination model and converting the paper type discrimination model into a binary file.

[0172] The process shown in Fig. 35 can be executed by running R on Linux using a CPU on which the Linux OS is installed. Note that the generation of the paper type discrimination model and the like may be executed using Python instead of R.

[0173] As shown in the upper part of Fig. 35, when training data is input, missing data processing, data analysis processing, etc. are performed, and then a discrimination model is generated using random forest and output as a Tree information file representing the node state. Next, as shown in the lower part of Fig. 35, a paper type discrimination algorithm (binary file) that can be incorporated into an executable file is generated from the output Tree information file, and output as a Tree binary file. The paper type discrimination algorithm (binary file) is written to a rewritable area (segment) dedicated to the paper type discrimination algorithm, which is provided in advance in the flash memory area inside the CPU.

[0174] (Setting the execution conditions for updating the paper type discrimination model) Fig. 36A is an example of a screen for accepting settings of execution conditions for updating the paper type discrimination model, and Fig. 36B is an example of a screen for accepting selection of the paper type discrimination model.

[0175] The image forming apparatus 10 displays, for example, a screen as shown in FIG. 36A on the operation panel 18, and receives settings of execution conditions for updating the paper type discrimination model from the user.

[0176] In the screen of FIG. 36A, the user can select "standby state" or "designated time" for the update timing (update time), which is an execution condition, by operating the touch panel button. When "designated time" is selected as the update timing, the user designates the update time (time at which the update is performed) by pressing the up and down buttons "△" and "▽" on the touch panel button, or by using a numeric keypad or soft key (not shown). As the update time, a day of the week, a date, or the like may be set in addition to the time. Alternatively, the elapsed time from the previous update may be set as the update time. In addition, the user can set the number of processes related to the number of updates N1 of the teacher data, which is an execution condition, in the same setting method as the setting of the update time. In the example of FIG. 36A, the predetermined number is set to 10, and when 10 sets of new teacher data are added, the execution condition related to the number of updates of the teacher data is satisfied. In addition, the user can specify whether to update the paper type discrimination model or not, or to select the previous version, by selecting a button in the paper type discrimination model column on the screen of FIG. 36A. When the "Select Previous Version" button is operated on the screen of FIG. 36A, a screen such as that shown in FIG. 36B is displayed. On the screen of FIG. 36B, the user can select whether to use the previous version of the paper type discrimination model, the next version of the paper type discrimination model, or the default model. Each time the "Select Previous Model" touch panel button is pressed, the user can go back from the currently selected version of the paper type discrimination model to the previous version of the paper type discrimination model. Conversely, each time the "Select Next Model" touch panel button is pressed, the paper type discrimination model can be returned to the most recent updated version. In this case as well, although not shown, a display is provided that displays the date when the version of the paper type discrimination model was registered, to allow the user to recognize the version of the paper type discrimination model selected by the user.In addition, if the touch panel button "Select previous model" or "Select next model" is pressed and held down and there are no more paper type discrimination model options available, a warning may be displayed indicating that there are no more options available, or a warning may be displayed indicating that the default model paper type discrimination model will be selected, or that the latest version of the paper type discrimination model will be selected, etc.

[0177] According to the second embodiment, the discrimination process can be executed using the latest updated paper type discrimination algorithm. Also, by incorporating the paper type discrimination model into the program as a paper type discrimination algorithm, the time required for the discrimination process can be shortened. Therefore, the paper setting process according to this embodiment can be included in the work sequence of the site where the image forming apparatus 10 is used. On the other hand, by performing the generation of the paper type discrimination model, the conversion to the paper type discrimination algorithm, and the incorporation process into the program, which take time, when the image forming apparatus 10 is not in a standby state or an operating state such as printing and adjustment, the impact on the normal use of the image forming apparatus 10 can be minimized.

[0178] Furthermore, according to the second embodiment, a paper type discrimination model is generated for each image forming apparatus 10, so that an appropriate paper type discrimination model according to the usage status of each image forming apparatus 10 can be generated.

[0179] Third embodiment In the first and second embodiments, examples have been described in which the image forming apparatus 10 performs paper discrimination processing, machine learning processing, and the like by itself. In the third embodiment, an example will be described in which each image forming apparatus 10 performs various processes in cooperation with a server 80 connected via a network L. The third embodiment will be described separately as Embodiment 3-1 to Embodiment 3-3 depending on the function executed by each component.

[0180] (3-1 embodiment) In the 3-1 embodiment, a teacher database and a machine learning function are provided in the server 80, and a paper type discrimination model and a paper type discrimination algorithm, which are trained models, are generated in the server 80. The generated paper type discrimination algorithm is transmitted from the server 80 to the image forming apparatus 10, and the image forming apparatus 10 executes a paper discrimination process using the paper type discrimination algorithm. The configuration and process of the 3-1 embodiment will be described in detail below.

[0181] The hardware configuration of the image forming apparatus 10 of the 3-1 embodiment is similar to that of the image forming apparatus 10 of the first and second embodiments, and therefore a detailed description thereof will be omitted.

[0182] (Server 80) FIG. 37 is a block diagram showing the configuration of the server.

[0183] As shown in FIG. 37, the server 80 includes a control unit 81, a communication unit 82, and a storage unit 83 (also referred to as a database), and each component is connected to each other via a bus so that they can communicate with each other. The server 80 may be provided in the same place as the image forming apparatus 10, or in a remote location and connected via a network. For example, the server 80 may be a cloud server that is virtually constructed by a plurality of servers arranged on a network such as the Internet. In this embodiment, the server 80 or the control unit 81 cooperates with the image forming apparatus 10 to configure a discrimination system. Also, in this embodiment, the image forming apparatus 10 or the control unit 11 (overall control unit 110) functions as a discrimination device.

[0184] The control unit 81 includes a CPU, a RAM, a ROM, etc., similar to the configuration of the control unit 11 of the image forming apparatus 10.

[0185] The communication unit 82 is capable of wired communication according to standards such as Ethernet (registered trademark), and wireless communication using standards such as Wi-Fi.

[0186] The storage unit 83 stores various information such as setting information required when performing image formation in each image forming apparatus 10, a database of teacher data used in machine learning, and the like.

[0187] Fig. 38 is a block diagram showing the functional configuration of a control unit of a server according to the embodiment 3-1 Fig. 39 is a block diagram showing the functional configuration of an overall control unit of an image forming apparatus according to the embodiment 3-1.

[0188] As shown in Fig. 38, the control unit 81 of the server 80 functions as an acquisition unit 811 and a learning unit 812. The acquisition unit 811 acquires measurement value information, paper type information, and the like, which serve as training data, from the image forming apparatus 10, etc. The learning unit 812 performs machine learning based on the training data, and generates a paper type discrimination model and a paper type discrimination algorithm, which are trained models. The function of the learning unit 812 is similar to the processing in the learning unit 113 of the overall control unit 110 of the image forming apparatus 10 described in Fig. 32, and therefore a detailed description thereof will be omitted.

[0189] 39, the overall control unit 110 of the image forming apparatus 10 functions as an acquisition unit 111, a discrimination unit 112, an output unit 114, and a reception unit 115. The functions of the acquisition unit 111 and the discrimination unit 112 are similar to those of the image forming apparatus 10 of the first and second embodiments, and therefore detailed description thereof will be omitted. The output unit 114 outputs the discrimination result by the discrimination unit 112 by displaying it on the operation panel 18, etc. The reception unit 115 receives an instruction to change the discrimination result output by the output unit 114.

[0190] The processes executed by each component will be described below.

[0191] (Transmission of change information in the image forming device to the server) 40 is a flowchart showing a process of transmitting modification information executed in an image forming apparatus according to the third embodiment. The following will be described with reference to FIG.

[0192] (Step S301) The overall control unit 110 of the image forming apparatus 10 judges whether or not the result of the judgment by the judgment unit 112 regarding the paper type, etc. has been changed by the user.

[0193] If no change has been made (step S301: NO), overall control unit 110 ends the process.

[0194] If it has been changed (step S301: YES), overall control unit 110 proceeds to the process of step S302.

[0195] (Step S302) The overall control unit 110 sets the changed paper type etc. as the paper setting to be used for image formation.

[0196] (Step S303) The overall control unit 110 judges whether or not a paper type other than the four types has been selected (set). In this embodiment, gloss coated paper, matte coated paper, plain paper, and fine paper are set as the four types of paper.

[0197] If a paper type other than the four types is selected (step S303: YES), for example, if the type is changed to "color paper," "envelope," or "embossed paper" in the example of FIG. 29, overall control unit 110 ends the process.

[0198] If four paper types are selected (step S303: NO), the overall control unit 110 proceeds to step S304 and executes the process of adding the paper types to the teacher database. Note that, in the present embodiment, an example of determining whether the changed paper type is one of the four predetermined types has been described, but the number and kinds of the predetermined paper types are not limited to this.

[0199] (Step S304) The overall control unit 110 acquires the corresponding measurement information. The measurement information is information on the measurement values ​​of the paper thickness, basis weight (density), paper surface properties, etc., which are input to the trained model in order to determine the paper type, etc., in the discrimination unit 112.

[0200] (Step S305) The overall control unit 110 transmits the paper type information changed by the user and the corresponding measurement value information to the server 80 as new teacher data (additional teacher data) to be added to the teacher database. At this time, the overall control unit 110 may transmit identification information such as a serial number for identifying the image forming device 10 to the server 80 together with the additional teacher data. This allows the server 80 to determine which image forming device 10 transmitted the additional teacher data and manage the addition history of the teacher data for each image forming device 10. As a result, the server 80 can also create a teacher database for each image forming device 10 and generate a paper type discrimination model for each image forming device 10.

[0201] (Adding changes made on the server to teacher data) 41 is a flowchart showing the process of adding teacher data executed in the server according to the embodiment 3-1. The following will be described with reference to FIG.

[0202] (Step S311) The control unit 81 of the server 80 determines whether or not there is additional teacher data transmitted from the image forming device 10.

[0203] If there is no additional teacher data (step S311: NO), the control unit 81 ends the process.

[0204] If there is additional teacher data (step S311: YES), the control unit 81 proceeds to the process of step S312.

[0205] (Steps S312 and S313) The control unit 81 reads the address information of the teacher database stored in the storage unit 83, and acquires the final address of the teacher database.

[0206] (Step S314) The control unit 81 writes additional teaching data, which is the information on the paper type changed by the user and the corresponding measurement value information, to the address next to the last address of the teaching database.

[0207] (Step S315) The control unit 81 counts up (increments) the number of updates N1, which is the number of times teacher data has been added, by one, stores it in the storage unit 83, and ends the process.

[0208] (Generation of paper type detection models and paper type detection algorithms on the server) 42 is a flowchart showing a process for generating a paper type discrimination model and a paper type discrimination algorithm executed in the server according to the third embodiment. The following will be described with reference to FIG.

[0209] (Step S321) The control unit 81 of the server 80 reads out the update count N1 of the teacher data from the storage unit 83.

[0210] (Step S322) The control unit 81 determines whether the number of updates N1 of the teacher data is equal to or greater than a predetermined number of times set in advance in the storage unit 83 as an execution condition for updating the paper type discrimination model. The execution condition for updating the paper type discrimination model is the same as that in the second embodiment, and is set by the user via the image forming apparatus 10 or a setting PC (not shown) or the like, and is stored in the storage unit 83 of the server 80.

[0211] If the number of times N1 the teacher data has been updated is not equal to or greater than the predetermined number of times (step S322: NO), the control unit 81 ends the process.

[0212] If the number of times N1 the teacher data has been updated is equal to or greater than the predetermined number of times (step S322: YES), the control unit 81 proceeds to the process of step S323.

[0213] (Step S323) The control unit 81 sets the number of times N1 the teacher data has been updated to zero.

[0214] (Step S324) The control unit 81 generates a paper type discrimination model (FIG. 31) using a random forest based on the teacher database (FIG. 30) stored in the storage unit 83. Note that the control unit 81 may generate the paper type discrimination model at a timing that is preset as an execution condition.

[0215] (Step S325) The control unit 81 generates a paper type discrimination algorithm that can be executed as a program based on the generated paper type discrimination model. The processes of steps S325 and S326 are similar to the processes of steps S219 and S220 in Fig. 34, and therefore detailed explanations are omitted. Note that the control unit 81 may generate the paper type discrimination algorithm at a timing that is preset as an execution condition.

[0216] (Step S326) The control unit 81 stores the generated paper type discrimination algorithm together with version information in the storage unit 83 as a reference algorithm.

[0217] (Update of paper type discrimination algorithm in image forming device) 43 is a flowchart showing the process of updating the paper type discrimination algorithm executed in the image forming apparatus according to the third embodiment. The following will be described with reference to FIG.

[0218] (Step S331) The overall control unit 110 of the image forming apparatus 10 accesses the server 80 to obtain update information for the paper type discrimination algorithm. For example, the overall control unit 110 obtains version information of the latest paper type discrimination algorithm stored in the server 80.

[0219] (Step S332) The overall control unit 110 compares the latest version information of the paper type discrimination algorithm acquired from the server 80 with the version information of the paper type discrimination algorithm used by the own machine to determine whether the paper type discrimination algorithm has been updated.

[0220] If not updated (step S332: NO), overall control unit 110 ends the process.

[0221] If it has been updated (step S332: YES), overall control unit 110 proceeds to the process of step S333.

[0222] (Step S333) The overall control unit 110 acquires information on the execution conditions for updating the paper type discrimination model from the server 80, the storage unit 12 of the image forming apparatus 10, or the like.

[0223] (Step S334) General control unit 110 judges whether the update timing preset as an execution condition is the "standby state".

[0224] If the update timing preset as the execution condition is the "standby state" (step S334: YES), general control unit 110 proceeds to the process of step S335.

[0225] If the update timing preset as the execution condition is not the "standby state" (step S334: NO), general control unit 110 proceeds to the process of step S336.

[0226] (Step S335) General control unit 110 determines whether image forming apparatus 10 is in a standby state.

[0227] If the image forming apparatus 10 is not in a standby state (step S335: NO), the overall control unit 110 continues the process of monitoring the state of the image forming apparatus 10 until the image forming apparatus 10 is in a standby state. Alternatively, the overall control unit 110 may end the process once and execute the process of FIG. 43 again at a predetermined timing.

[0228] If it is in the standby state (step S335: YES), overall control unit 110 proceeds to the process of step S338.

[0229] (Step S336) General control unit 110 judges whether or not the update timing preset as an execution condition is a "designated time".

[0230] If the update timing preset as the execution condition is the "designated time" (step S336: YES), general control unit 110 proceeds to the process of step S337.

[0231] If the update timing preset as the execution condition is not the "designated time" (step S336: NO), the overall control unit 110 ends the process assuming that the update execution condition is not set. Alternatively, the overall control unit 110 may determine the update execution condition based on a predetermined default setting or the like.

[0232] (Step S337) General control section 110 refers to the date and time managed in date and time management section 121, and determines whether or not a preset designated time has arrived.

[0233] If the specified time has not arrived (step S337: NO), the overall control unit 110 continues the process of monitoring the time when the specified time arrives. Alternatively, the overall control unit 110 may end the process once and execute the process of FIG. 43 again at a predetermined timing.

[0234] If the designated time has arrived (step S337: YES), overall control unit 110 proceeds to the process of step S338.

[0235] (Step S338) The overall control unit 110 prohibits the reception of instructions from the user or the like regarding the printing operation or adjustment operation of the image forming apparatus 10.

[0236] (Step S339) The overall control unit 110 requests the server 80 to obtain the latest paper type discrimination algorithm.

[0237] (Step S340) The overall control unit 110 writes the latest paper type discrimination algorithm acquired from the server 80 into a designated segment of the flash memory of the CPU of the overall control unit 110 .

[0238] (Step S341) General control unit 110 permits (cancels) reception of instructions from a user or the like regarding the printing operation or adjustment operation of image forming apparatus 10 that was prohibited in step S338.

[0239] According to the 3-1 embodiment, the time-consuming process of generating a paper type discrimination model is executed by the server 80, thereby shortening the blocking time of the print process of the image forming device 10, which is accompanied by the process of generating a paper type discrimination model. Furthermore, by having the server 80 generate and store the latest updated paper type discrimination algorithm, the latest paper type discrimination algorithm can be shared by multiple image forming devices 10 connected to the server 80. Furthermore, by creating a teacher database for each image forming device 10 in the server 80, it is also possible to generate a paper type discrimination algorithm according to the usage status, etc. of each image forming device 10.

[0240] (Embodiment 3-2) In the third embodiment, a paper type discrimination algorithm, which is a trained model, is implemented in the server 80, and the paper discrimination process is executed in the server 80. The configuration and process of the third embodiment will be described in detail below.

[0241] The hardware configurations of the image forming apparatus 10 and the server 80 of the 3-2 embodiment are similar to those of the 3-1 embodiment.

[0242] FIG. 44A is a block diagram showing a functional configuration of a control unit of a server according to embodiment 3-2.

[0243] As shown in Fig. 44A, the control unit 81 of the server 80 functions as an acquisition unit 811 and a discrimination unit 813. The function of the acquisition unit 811 is similar to that of the acquisition unit 811 of the server 80 of the third embodiment described in Fig. 37, and therefore a detailed description thereof will be omitted. The function of the discrimination unit 813 is similar to the processing in the discrimination unit 112 of the overall control unit 110 of the image forming apparatus 10 described in Fig. 18B, and therefore a detailed description thereof will be omitted. In this embodiment, the server 80 or the control unit 81 functions as a discrimination device.

[0244] FIG. 44B is a block diagram showing the functional configuration of an overall control unit of the image forming apparatus according to the third embodiment.

[0245] 44B, the overall control unit 110 of the image forming apparatus 10 functions as an acquisition unit 111, an output unit 114, and a reception unit 115. The functions of the acquisition unit 111, the output unit 114, and the reception unit 115 are similar to those of the image forming apparatus 10 of the 3-1 embodiment, and therefore detailed description thereof will be omitted.

[0246] The processes executed by each component will be described below.

[0247] (Paper setting process in image forming device) 45 is a flowchart showing the paper setting process (corresponding to step S10 in FIG. 19) executed in the image forming apparatus according to the third embodiment. The following will be described with reference to FIG.

[0248] (Steps S350 and S351) The transport control / image formation control unit 160 of the control unit 11 causes the opening / closing mechanism 65 to open the shutter 651 , and then feeds the paper S from the selected tray and transports it to the transport path 143 .

[0249] (Step S352a) The media sensor control unit 130 detects the thickness of the paper S using the paper thickness sensor 40, and passes the detection data (measurement value 1) to the overall control unit 110.

[0250] (Steps S352b and S352c) The media sensor control unit 130 detects the basis weight and surface properties of the paper using the basis weight sensor 50 and the surface property sensor 60, and passes the detection data (measured values ​​2 and 3) to the overall control unit 110. The processing of steps S352b and S352c is similar to steps S102b and S102c in FIG. 20, and therefore detailed description thereof will be omitted.

[0251] (Step S353) General control unit 110 judges whether a predetermined number N of measurements have been performed.

[0252] If the predetermined number N of measurements have not been performed (step S353: NO), general control unit 110 repeats the processes of steps S352b and S352c.

[0253] If the predetermined number N of measurements have been performed (step S353: YES), general control unit 110 proceeds to the process of step S354.

[0254] (Step S354) When the rear end of the paper S passes the media sensor 15, specifically, when the paper sensor (not shown) detects that the rear end of the paper S has left the transport rollers 1434 (YES), the process proceeds to step S355.

[0255] (Step S355) Here, the control unit 11 closes the shutter 651 using the opening / closing mechanism 65 to keep it in a closed state. This prevents the surface property sensor 60 from being soiled by paper powder or the like caused by the paper S being transported when the media sensor 15 is not performing a measurement.

[0256] (Step S356) Overall control unit 110 performs averaging processing on the detection data (measurement values ​​2 and 3) of basis weight and surface property for N times acquired in steps S352b and S352c. Overall control unit 110 stores the information acquired in the processing of steps S352a to S356 in memory unit 12 as measurement value information.

[0257] (Step S357) The overall control unit 110 transmits the measurement value information to the server 80 .

[0258] (Step S358) The overall control unit 110 acquires from the server 80 the results of the determination of the paper type / basis weight classification and the like determined by the server 80 based on the measurement value information. The determination process in the server 80 will be described later.

[0259] (Step S359) The overall control unit 110 obtains from the server 80 the result of the determination of the registered profile made by the server 80 based on the measurement value information.

[0260] (Step S360) The overall control unit 110 displays the determination results acquired from the server 80 on the operation panel 18, and accepts instructions from the user such as application or modification of paper types, registered profiles, control parameters, and the like.

[0261] (Step S361) The overall control unit 110 determines whether or not the information such as the paper type acquired as the determination result has been changed by the user.

[0262] If the obtained information is accepted without being changed as a result of the determination (step S361: NO), overall control unit 110 proceeds to the process of step S362.

[0263] If the determined result is that the acquired information has been changed (step S361: YES), overall control unit 110 proceeds to the process of step S363.

[0264] (Step S362) The overall control unit 110 sets the acquired information such as the paper type in the selected tray of the image forming apparatus 10 .

[0265] (Step S363) The overall control unit 110 sets the changed information such as the paper type in the selected tray of the image forming apparatus 10, and transmits an instruction to the server 80 to register the changed information.

[0266] (Step S364) The overall control unit 110 sets the printing conditions for the selected tray to printing conditions that correspond to the set information such as paper type.

[0267] (Determination process on the server) 46 is a flowchart showing a process of determining the paper type and the like executed by the server according to the third embodiment. The following will be described with reference to FIG.

[0268] (Step S371) The control unit 81 of the server 80 acquires the measurement value information transmitted from the image forming apparatus 10 in the process of step S357 in FIG.

[0269] (Step S372) The control unit 81 determines the paper type and the basis weight class based on the measurement value information. The process of step S372 is similar to the paper type determination and the basis weight class determination process of step S107 in Fig. 20, so a detailed description thereof will be omitted.

[0270] (Step S373) The control unit 81 reads out a preset profile, which is a registered profile stored in advance in the storage unit 83.

[0271] (Step S374) The control unit 81 determines candidates for the preset profile to be used based on the paper type and basis weight category determined in step S372.

[0272] (Step S375) The control unit 81 reads out a user profile, which is a registered profile that is registered by a user and stored in the storage unit 83.

[0273] (Step S376) The control unit 81 determines candidates for the user profile to be used based on the paper type and basis weight category determined in step S372. The process of determining candidates for the profile to be used will be described in detail later with reference to FIG.

[0274] (Step S377) The control unit 81 transmits to the image forming apparatus 10 information indicating the paper type and basis weight category determined in step S372, the preset profile determined in step S374, and the user profile candidates determined in step S376.

[0275] (Registration of change information on the server) 47 is a flowchart showing a registration process of change information executed in the server according to the embodiment 3-2. The following will be described with reference to FIG.

[0276] (Step S381) The control unit 81 of the server 80 determines whether or not an instruction to register the change information transmitted from the image forming apparatus 10 in the process of step S363 in FIG. 45 has been received.

[0277] If not received (step S381: NO), the control unit 81 ends the process.

[0278] If it has been received (step S381: YES), the control unit 81 proceeds to the process of step S382.

[0279] (Steps S382, S383) The control unit 81 acquires change information including information such as the changed paper type, the registered profile, and the control parameters from the image forming apparatus 10, and registers the acquired change information in the storage unit 83.

[0280] According to embodiment 3-2, the discrimination process is executed on the server 80, so that the discrimination process can be standardized among multiple image forming devices 10 connected via a network, and the same discrimination result can be obtained even for different image forming devices 10.

[0281] In addition, setting information such as user profiles and control parameters stored for each image forming device 10 can also be stored in the server 80 for each image forming device 10, so that, for example, the user profile and control parameters of a particular user registered in one image forming device 10 can also be used in other image forming devices 10.

[0282] (Embodiment 3-3) In the 3-3 embodiment, similar to the 3-1 embodiment, a teacher database and a machine learning function are provided in the server 80, and a paper type discrimination model and a paper type discrimination algorithm, which are trained models, are generated in the server 80. Furthermore, in the 3-3 embodiment, similar to the 3-2 embodiment, the generated paper type discrimination algorithm is implemented in the server 80, and a paper discrimination process using the paper type discrimination algorithm is executed in the server 80. The configuration and process of the 3-3 embodiment will be described in detail below.

[0283] The hardware configurations of the image forming apparatus 10 and the server 80 of the 3-3 embodiment are similar to those of the 3-1 and 3-2 embodiments.

[0284] FIG. 48A is a block diagram showing a functional configuration of a control unit of a server according to embodiment 3-3.

[0285] As shown in Fig. 48A, the control unit 81 of the server 80 functions as an acquisition unit 811, a learning unit 812, and a discrimination unit 813. The functions of the acquisition unit 811 and the learning unit 812 are similar to those of the acquisition unit 811 of the server 80 of the 3-1 embodiment described in Fig. 37, and therefore detailed description thereof will be omitted. In addition, the function of the discrimination unit 813 is similar to that of the discrimination unit 813 of the server 80 of the 3-2 embodiment described in Fig. 44, and therefore detailed description thereof will be omitted. In this embodiment, the server 80 or the control unit 81 functions as a discrimination device or discrimination system.

[0286] FIG. 48B is a block diagram showing the functional configuration of an overall control unit of an image forming apparatus according to the third embodiment.

[0287] 48B, the overall control unit 110 of the image forming apparatus 10 functions as an acquisition unit 111, an output unit 114, and a reception unit 115. The functions of the acquisition unit 111, the output unit 114, and the reception unit 115 are similar to those of the image forming apparatus 10 of the third embodiment and the third embodiment, and therefore detailed description thereof will be omitted.

[0288] The processes executed by each component will be described below.

[0289] (Paper setting process in image forming device) The paper setting process executed in the image forming apparatus 10 of the 3-3 embodiment is similar to the paper setting process in the image forming apparatus 10 of the 3-2 embodiment described with reference to FIG.

[0290] (Registering changes on the server and adding them to teacher data) 49 is a flowchart showing a registration process of change information executed in a server according to the embodiment 3-3. The following will be described with reference to FIG.

[0291] (Step S391) The control unit 81 of the server 80 determines whether or not an instruction to register the change information transmitted from the image forming apparatus 10 in the process of step S363 in FIG. 45 has been received.

[0292] If not received (step S391: NO), the control unit 81 proceeds to the process of step S394.

[0293] If it has been received (step S391: YES), the control unit 81 proceeds to the process of step S392.

[0294] (Steps S392 and S393) The control unit 81 acquires change information including information such as the changed paper type, the registered profile, and the control parameters from the image forming apparatus 10, and registers the acquired change information in the storage unit 83.

[0295] (Step S394) The control unit 81 determines whether or not an instruction regarding addition to the teacher data has been received from the image forming apparatus 10.

[0296] If not received (step S394: NO), the control unit 81 ends the process.

[0297] If it has been received (step S394: YES), the control unit 81 proceeds to the process of step S395.

[0298] (Steps S395 and S396) The control unit 81 acquires the changed paper type and the corresponding measurement value information from the image forming apparatus 10, associates the information with the changed paper type, and adds the associated information to the teacher database in the storage unit 83. The process of adding the information to the teacher database is similar to the process of adding the information to the teacher database in the server 80 of the 3-1 embodiment described in FIG.

[0299] (Generation of paper type detection models and paper type detection algorithms on the server) 50 is a flowchart showing a process for generating and updating a paper type discrimination model and a paper type discrimination algorithm executed in a server according to the third embodiment. The following will be described with reference to FIG.

[0300] (Steps S3101 to S3105) The processing in steps S3101 to S3105 is similar to the processing in steps S321 to S325 in the server 80 in the 3-1 embodiment described with reference to FIG. 42, and therefore a detailed description thereof will be omitted.

[0301] (Step S3106) The control unit 81 of the server 80 writes the executable paper type discrimination algorithm generated in the process of step S3105 to a designated segment of the flash memory of the CPU of the control unit 81, and ends the process.

[0302] According to the third embodiment, the generation and update processes of the paper type discrimination model, which take a long time to process, are executed in the server 80, so that the blocking time of the print process of the image forming device 10, which is accompanied by the generation and update processes of the paper type discrimination model, is minimized. In addition, the server 80 generates and executes the latest updated paper type discrimination algorithm, so that the multiple image forming devices 10 connected to the server 80 can share the latest paper type discrimination algorithm. In addition, the server 80 can generate and execute a paper type discrimination algorithm according to the usage status of each image forming device 10 by creating a teacher database for each image forming device 10. For example, the paper type discrimination algorithm may be provided for each individual image forming device 10 of the same model. In this case, a paper type discrimination algorithm is provided for each image forming device 10a, 10b, 10c, etc., which are different machines of the same model of the image forming device 10. In addition, the paper type discrimination algorithm may be provided for each model of the image forming device 10. In this case, a paper type discrimination algorithm is provided for each image forming device 10 of different models, such as image forming device A, image forming device B, image forming device C, etc., which are different models of the image forming device 10. Alternatively, the paper type discrimination algorithm may be provided for each model and each individual image forming apparatus 10. In this case, a paper type discrimination algorithm is provided for each of different models of image forming apparatus 10, such as image forming apparatus A, image forming apparatus B, image forming apparatus C, etc., such as image forming apparatus A-10a, image forming apparatus A-10b, image forming apparatus A-10c, etc., image forming apparatus B-10a', image forming apparatus B-10b', image forming apparatus B-10c', etc., image forming apparatus C-10a", image forming apparatus C-10b", image forming apparatus C-10c", etc.

[0303] (A variant using a neural network as the trained model) In the above embodiment, an example has been described in which a paper type discrimination model, which is a trained model, is generated using a random forest model. However, the paper type discrimination model may be generated using other machine learning models such as a neural network.

[0304] FIG. 51 is a diagram showing an example of discriminating the paper type using a trained model of a neural network that has been machine-learned using training data.

[0305] As shown in FIG. 51, a deep neural network (DNN) is configured as a neural network having an input layer, an output layer, and multiple hidden layers (intermediate layers), and deep learning using teacher data is performed. In the example of the DNN shown in FIG. 51, data obtained by averaging the amount of specular reflection light and the amount of diffuse reflection light measured by the surface property sensor 60, and data obtained by averaging the values ​​related to basis weight and paper thickness measured by the basis weight sensor 50 and the paper thickness sensor 40 are input to the input layer. Furthermore, data obtained by performing a predetermined pre-processing on a surface image of a paper captured by a camera (described later) is also input to the input layer. Data related to the type of paper to be identified, such as gloss coated paper, matte coated paper, fine paper, and plain paper, is output from the output layer. A paper type identification model is generated by machine learning the DNN shown in FIG. 51 using the teacher data shown in FIG. 30. Note that the machine learning process of the neural network and DNN using teacher data and the identification process using a trained model are publicly known, so detailed explanations will be omitted.

[0306] Fig. 52 is a diagram illustrating an example of an imaging mechanism for acquiring surface images. Fig. 53 is a diagram illustrating surface images of gloss coated paper, matte coated paper, fine paper, and plain paper acquired by the imaging mechanism, and their respective characteristics.

[0307] As shown in FIG. 52, the photographing mechanism 1500 has a camera 1501, an LED 1502, and a reflector 1503. The photographing mechanism 1500 is provided, for example, on the transport path of the paper S in the image forming apparatus 10. The camera 1501 is, for example, a CMS color camera module provided with a close-up macro lens, and has a pixel size of 2 μm to 10 μm. The transported paper S is placed at a predetermined paper position by a paper pressing mechanism. The light output from the LED 1502 is reflected by the reflector 1503 and irradiates the surface of the paper S placed at the predetermined paper position. The LED 1502 uses a white-emitting LED. The paper surface image difference can be detected when the irradiation angle for the paper S is set in the range of 95° to 80°. The closer to 95°, the more oblique the light is applied, so it is easier to detect the difference in unevenness on the surface of the paper S, but as a result of verification, 90° is preferable. The distance and depth of the subject can be set by the focal length and brightness of the lens, but in order to accurately read the image of the paper surface condition, the paper S is stopped and fixed with a push-up plate in the same way as the surface property sensor 60 and read. The camera 1501 takes an image of the surface of the paper S where the light output from the LED 1502 is irradiated, and obtains surface images of each type of paper S as shown in FIG. 53. As shown in FIG. 53, the smoothness of the paper surface decreases, the surface becomes rougher, and the image density increases in the order of gloss coated paper, matte coated paper, fine paper, and plain paper. In the above, the LED 1502 is an LED that emits white light in order to read the color of the paper surface as well, but an LED that emits light of a color other than white or infrared light may be used to extract the characteristics of the image of the paper surface. Also, a white-light-emitting LED may be used, and a cut filter that passes only a specific light wavelength may be attached in front of the camera lens.

[0308] Data representing the surface image acquired by the imaging mechanism 1500 is subjected to a predetermined pre-processing by, for example, the control unit 11 (overall control unit 110) and then added to a teacher database stored in the image forming apparatus 10 or the server 80. As the predetermined pre-processing, for example, in order to acquire the characteristic amount of unevenness on the paper surface, an integration process of a predetermined frequency band when the surface image is frequency analyzed may be executed. Alternatively, as the predetermined pre-processing, a process of outputting the average gradation (overall scattered light amount) and the standard deviation (shade variation) of one line of the image may be executed. As another predetermined pre-processing, a binarization process using a threshold value that reveals the characteristic of unevenness on the paper surface may be executed.

[0309] In this way, by generating a paper type discrimination model using a neural network (deep neural network), it becomes easy to introduce a surface image as an input factor for machine learning. This allows the characteristics and state of the unevenness of the paper surface, which is expressed as the amount of variation in density in the surface image, to be used as parameters for paper type discrimination, so that paper type discrimination can be performed with high accuracy. Note that when performing machine learning, it takes time to tune (adjust) the weights of each layer of the neural network. Therefore, as in each of the above embodiments, the effect of the measures to suppress the impact on normal use of the image forming apparatus 10 by controlling whether or not to perform the process of generating or updating a trained model, the timing of performing the process, the configuration for performing the process, etc., is more pronounced.

[0310] (Regarding the first and second basis weights) A method for acquiring the first and second basis weights and an example of using the basis weight difference (first basis weight - second basis weight) to determine the paper type will be described below.

[0311] First, the configuration of the basis weight sensor 50 will be described in detail.

[0312] (Configuration of basis weight sensor 50) 54 is a schematic diagram showing the configuration of the basis weight sensor 50. As described above, the basis weight sensor 50 is a transmission type optical sensor that detects the basis weight of the paper, and is equipped with a light emitting section and a light receiving section, and measures the basis weight based on the amount of attenuation of light passing through the paper S.

[0313] As shown in FIG. 54, the basis weight sensor 50 includes a plurality of light-emitting units 51 and a single light-receiving unit 52. The light-emitting unit 51 includes a first light-emitting unit 51a, a second light-emitting unit 51b, and a third light-emitting unit 51c. The first, second, and third light-emitting units respectively irradiate the irradiation area with the first, second, and third irradiation light. This irradiation area (second irradiation area) is an inner area of ​​the opening a12 when viewed from the Z' direction. The opening a12 is provided in the upper guide 1431. The lower guide 1432 also has an opening a22 at a position opposite the opening a12. The openings a22 and a12 have the same shape, for example, a rectangle. In order to prevent foreign matter such as paper powder from the paper S passing through the transport path 143 from adhering to the openings a22 and a12, transparent sheets 54a and 54b made of PET or the like that transmit the wavelengths of the irradiation light are attached.

[0314] The first light-emitting unit 51a emits a first irradiation light having a first wavelength. The first wavelength is, for example, a near-infrared wavelength longer than the wavelength of visible light. More specifically, the first wavelength includes, for example, a wavelength between 750 nm and 900 nm. The second light-emitting unit 51b emits a second irradiation light having a second wavelength. The second wavelength is, for example, a wavelength of blue light included in visible light. More specifically, the second wavelength includes, for example, a wavelength between 400 nm and 470 nm. The first light-emitting unit 51a and the second light-emitting unit 51b are both disposed on the opposite side of the transport path 143 from the light-receiving unit 52, and the third light-emitting unit 51c is disposed on the same side as the light-receiving unit 52 and in the vicinity of the light-receiving unit 52. The third light-emitting unit 51c emits a third irradiation light having a third wavelength toward the irradiation area (opening a12). The third wavelength is, for example, a wavelength of green light among visible light. More specifically, the third wavelength includes, for example, a wavelength between 495 nm and 570 nm. The third wavelength is a wavelength different from the first wavelength (for example, a wavelength between 750 nm and 900 nm) and the second wavelength (for example, 400 nm to 470 nm).

[0315] The third irradiation light is irradiated toward the transport path 143 inside the upper and lower guides 1431 and 1432. A reflecting section 53 is provided inside the lower guide 1432 provided near the first light-emitting section 51a and the second light-emitting section 51b. The reflecting section 53 is painted, for example, in green, which is the same color as the third irradiation light, and reflects the third irradiation light. Note that the reflecting section 53 does not reflect the first irradiation light (near-infrared rays) and the second irradiation light (blue light rays), which are not the same color.

[0316] In this embodiment, the control unit 11 controls the first light emitting unit 51a and the second light emitting unit 51b during measurement to irradiate the first irradiation light and the second irradiation light at different timings. The light receiving unit 52 receives the first irradiation light and the second irradiation light, detects the light amount of each irradiation light, and outputs the detected light amount of the first irradiation light and the second irradiation light to the control unit 11. Similarly, the first irradiation light and the second irradiation light are irradiated to the paper S transported to the position of the opening a12. The light receiving unit 52 receives the transmitted light (first transmitted light, second transmitted light) of the first irradiation light and the second irradiation light, detects the light amount of each irradiation light, and outputs the detected light amount of the first transmitted light and the second transmitted light to the control unit 11. That is, the light receiving unit 52 detects the first irradiation light and the second irradiation light when there is no paper S, and the first transmitted light and the second transmitted light when the paper S is at the opening a12.

[0317] Similarly, with respect to the third light-emitting unit 51c, the light-receiving unit 52 detects the first reflected light reflected by the reflecting unit 53 when no paper S is present, and the second reflected light reflected by the surface of the paper S when the paper S is in the opening a12.

[0318] The control unit 11 calculates a first transmittance by dividing the amount of the first transmitted light by the amount of the first irradiated light. Similarly, the control unit 11 calculates a second transmittance by dividing the amount of the second transmitted light by the amount of the second irradiated light. Then, the type of the paper S is determined based on the first and second transmittances and the determination criteria stored in the storage unit 12.

[0319] Furthermore, in addition to the first and second transmittances, the control unit 11 may divide the amount of the third reflected light by the amount of the first reflected light to calculate the reflectance, and may determine the type of the paper S by taking this reflectance into consideration. Note that, although the third light-emitting unit 51c and the reflecting unit 53 are provided in this embodiment, they may be omitted.

[0320] Next, the function of the basis weight sensor 50 will be described.

[0321] (Function of Grammage Sensor 50) FIG. 55 is a block diagram showing the functional configuration of the basis weight sensor and control unit, in which the third light-emitting unit 51c and the reflecting unit 53 are omitted.

[0322] Hereinafter, with reference to Fig. 55, the function of controlling the light-emitting unit 51 will be described first, and the functions executed thereafter will be described in order. The control unit 11 functions as an irradiation control unit 1100 that controls the timing at which the light-emitting unit 51 emits irradiation light. The irradiation control unit 1100 outputs an instruction signal to instruct the first light-emitting unit 51a and the second light-emitting unit 51b to emit irradiation light. The first light-emitting unit 51a and the second light-emitting unit 51b emit the first irradiation light and the second irradiation light at different timings.

[0323] The light receiving unit 52 receives the first irradiation light and the second irradiation light, detects the amount of light of each irradiation light, and outputs the detected amount of light of the first irradiation light and the second irradiation light to the control unit 11. When the paper S is transported, the light receiving unit 52 receives first transmitted light obtained by transmitting the first irradiation light through the paper S and second transmitted light obtained by transmitting the second irradiation light through the paper S, and outputs the detected amount of light of the first transmitted light and the second transmitted light to the control unit 11.

[0324] The control unit 11 functions as the transmittance calculation unit 1120 and calculates the transmittance based on the amount of irradiation light and the amount of transmitted light. The transmittance calculation unit 1120 calculates the first transmittance by dividing the amount of the first transmitted light by the amount of the first irradiation light. The transmittance calculation unit 1120 calculates the second transmittance by dividing the amount of the second transmitted light by the amount of the second irradiation light.

[0325] The control unit 11, as the type determination unit 1140, determines the basis weight derivation paper type, which is the type of the paper S for selecting a formula for converting the basis weight, based on the calculated first transmittance and second transmittance. More specifically, the type determination unit 1140 determines the basis weight derivation paper type of the paper S using the first transmittance and second transmittance, a predetermined determination criterion among a plurality of determination criteria 124 stored in the memory unit 12, and the basis weight threshold value 126. Alternatively, the basis weight derivation paper type used by the basis weight derivation unit 1160 of the paper S may be determined from the reflected light amount value detected by the surface property sensor 60.

[0326] The control unit 11, functioning as the basis weight derivation unit 1160, derives the first basis weight and the second basis weight as basis weights according to a determination criterion based on the basis weight derivation paper type of the determined sheet S. The process of acquiring the first basis weight and the second basis weight will be described in detail later.

[0327] The criteria for judgment will be explained below.

[0328] (Judgment criteria) The judgment criteria will be described with reference to Figs. 56 to 58. Fig. 56 is a diagram showing the judgment criteria that express the relationship between the first transmittance and the nominal basis weight. More specifically, Fig. 56 shows the relationship between the transmittance and the nominal basis weight obtained when a first irradiation light (near-infrared light) having a first wavelength is irradiated onto plain paper, coated paper, and recycled paper. The transmittance is a value obtained by an experiment. The nominal basis weight is the nominal basis weight of the paper S published by the manufacturer that produces the paper S, etc.

[0329] In Fig. 56, the index showing the relationship between the transmittance and the nominal basis weight of plain paper is indicated by a diamond, the index showing the relationship between the transmittance and the nominal basis weight of coated paper is indicated by a square, and the index showing the relationship between the transmittance and the nominal basis weight of recycled paper is indicated by a triangle. For example, the index 601 indicates that the transmittance calculated by experiment for a certain plain paper is about 25%, and the nominal basis weight of the plain paper used in the experiment is about 100 g / m 2 As another example, the indicator 602 indicates that the experimentally calculated transmittance of a certain coated paper is about 30%, and the nominal basis weight of the coated paper used in the experiment is about 100 g / m 2 The transmittance of the recycled paper was also calculated by experiment, and the nominal basis weight of the recycled paper used in the experiment was indicated by an index.

[0330] In the experiment, a plurality of indices for plain paper, a plurality of indices for coated paper, and a plurality of indices for recycled paper are calculated. Based on these indices, an approximation line using, for example, the least squares method is derived by a control unit (not shown) provided in a PC (Personal Computer) used in the experiment. More specifically, a judgment criterion 611 (dash-dotted line) for plain paper, a judgment criterion 612 (solid line) for coated paper, and a judgment criterion 613 (dashed line) for recycled paper are derived. A plurality of judgment criteria 124 including the judgment criteria 611 to 613 are stored in the memory unit 12.

[0331] Fig. 57 is a diagram showing the criteria for expressing the relationship between the second transmittance and the nominal basis weight. More specifically, Fig. 57 shows the relationship between the transmittance and the nominal basis weight obtained when the second irradiation light having a second wavelength (blue light included in visible light) is irradiated onto plain paper, coated paper, and recycled paper. In Fig. 57, the indicator showing the relationship between the transmittance and the nominal basis weight of plain paper is indicated by a diamond, the indicator showing the relationship between the transmittance and the nominal basis weight of coated paper is indicated by a square, and the indicator showing the relationship between the transmittance and the nominal basis weight of recycled paper is indicated by a triangle.

[0332] A plurality of indices are calculated for each of plain paper, coated paper, and recycled paper. Based on these indices, an approximation line is derived, for example, using the least squares method, by a control unit provided in a PC used in the experiment. More specifically, a judgment criterion 711 (dash-dotted line) for plain paper, a judgment criterion 712 (solid line) for coated paper, and a judgment criterion 713 (dashed line) for recycled paper are derived. The memory unit 12 can store a plurality of judgment criteria 124 including the judgment criteria 711 to 713.

[0333] The storage unit 12 may store the approximation line judgment criteria as a table, or may store a formula corresponding to the approximation line. For example, in the judgment criteria 611 to 613, the formula when the basis weight is y1 is the following formula (1).

[0334] y1=exp(b1×Log(x1)+b2)...(1) x1 represents the first transmittance, and b1 and b2 represent constants. The first transmittance used in formula (1) is a value obtained by dividing the transmittance by 100 (for example, when the transmittance is 25%, 25÷100=0.25). The control device 101 reads formula (1) stored in the storage unit 12, and derives the basis weight by substituting the transmittance into formula (1). Note that by changing at least one of the constants b1 and b2, a formula corresponding to any one of the judgment criteria 611, judgment criteria 612, and judgment criteria 613 is generated.

[0335] For example, in the criteria 711 to 713, when the basis weight is y2, the formula is the following formula (2).

[0336] y2=exp(a1×Log(x2) 2 +a2×Log(x2)+a3) (2) x2 represents the second transmittance, and a1, a2, and a3 represent constants. The second transmittance used in formula (2) is a value obtained by dividing the transmittance by 100 (for example, when the transmittance is 30%, 30÷100=0.3). The control device 101 reads formula (2) stored in the storage unit 12, and derives the basis weight by substituting the transmittance into formula (2). Note that by changing at least one of the constants a1, a2, and a3, a formula corresponding to any one of the judgment criteria 711, the judgment criteria 712, and the judgment criteria 713 is generated.

[0337] The first and second wavelengths are used when calculating the transmittance because the indexes constituting the approximation line in the judgment criteria have smaller variations at these wavelengths than at other wavelengths. More specifically, the indexes showing the relationship between the transmittance and the nominal basis weight at the first and second wavelengths are more likely to be present on the approximation line or closer to the approximation line than the indexes showing the relationship between the transmittance and the nominal basis weight at other wavelengths. The first and second wavelengths are wavelengths that have a higher correlation between the transmittance and the basis weight than other wavelengths. Therefore, the first irradiation light having the first wavelength and the second irradiation light having the second wavelength are used to calculate the transmittance to derive the basis weight.

[0338] FIG. 58 is a diagram showing the correlation between transmittance and nominal basis weight at wavelength. Referring to FIG. 58, the first wavelength corresponding to the wavelength of near infrared rays (750 nm to 900 nm) having a relatively long wavelength (nm) and the second wavelength corresponding to the wavelength of blue light (400 nm to 470 nm) of visible light have a coefficient of determination closer to "1" than other wavelengths (for example, 500 nm to 700 nm). When the coefficient of determination at a certain wavelength is closer to "1", the irradiated light having the wavelength has a high correlation between the transmittance and the nominal basis weight. When the correlation between the transmittance and the nominal basis weight is high, the variation of the multiple indices with respect to the approximation line is smaller than when the correlation is low. Referring to FIG. 58, since the coefficients of determination at the first wavelength and the second wavelength are closer to "1", the irradiated light having these wavelengths has a high correlation between the transmittance and the nominal basis weight. The values ​​showing the correlation between transmittance and basis weight at the wavelengths in Fig. 58 were obtained by the experiments described in Fig. 56 and Fig. 57. Also, Fig. 58 shows the correlation for plain paper, but other types of paper S (e.g., coated paper, etc.) also have a similar correlation.

[0339] The reason why the correlation differs depending on the wavelength is that the components constituting the paper S are different even if the type of paper S is the same. When the type of paper S is the same, the nominal basis weight is approximately the same basis weight, but when the components are different, the transmittance detected when the paper S is irradiated with irradiation light may differ. The paper S contains at least one component of, for example, calcium carbonate, kaolin, talc, and satin white. When the paper S contains these components, the index based on the first irradiation light having a first wavelength and the index based on the second irradiation light having a second wavelength have small variations from the approximation line, and the index based on the irradiation light of another wavelength has large variations from the approximation line.

[0340] The image forming apparatus 10 calculates the transmittance using irradiation light having a wavelength that has a high correlation between the transmittance and the nominal basis weight compared to other wavelengths, and derives the basis weight corresponding to the transmittance, thereby enabling the accurate grammage to be derived. Even if the basis weight is derived using irradiation light having a wavelength with a high correlation, the difference between the derived basis weight and the nominal basis weight may be extremely large depending on the components of the paper S. Therefore, a basis weight with a small difference from the nominal basis weight is derived based on each basis weight using multiple irradiation lights having a high correlation wavelength (for example, a first irradiation light having a first wavelength and a second irradiation light having a second wavelength). The process of deriving a basis weight with a small difference from the nominal basis weight will be described in detail later.

[0341] (About basis weight difference (1st basis weight - 2nd basis weight) and basis weight threshold) FIG. 59 is a diagram showing an index representing the correspondence between the transmittance and the basis weight difference and a basis weight threshold 126. More specifically, the index represents the correspondence between the transmittance and the basis weight difference in plain paper and the correspondence between the transmittance and the basis weight difference in coated paper. The transmittance and the basis weight difference are values ​​obtained by an experiment. The transmittance is a first transmittance calculated using, for example, a first irradiation light (near infrared) having a first wavelength. The basis weight difference is the difference between the first basis weight and the second basis weight, as described above. The first basis weight is a basis weight corresponding to the transmittance derived using the determination criterion 611 for plain paper at the first wavelength shown in FIG. 56. The second basis weight is a basis weight corresponding to the transmittance derived using the determination criterion 711 for plain paper at the second wavelength shown in FIG. 57.

[0342] In Figure 59, the index showing the relationship between the transmittance and the basis weight difference of plain paper is represented by a diamond. The index for plain paper is a basis weight difference corresponding to the transmittance of about -10 g / m 2 The plots are made at the above positions. The indicators showing the relationship between the transmittance and the basis weight difference of coated paper are shown as squares. The indicators for coated paper are those where the basis weight difference corresponding to the transmittance is approximately -15g / m 2 The paper is plotted at the following positions. A threshold value is set to distinguish between plain paper and coated paper based on the basis weight difference between the plain paper and coated paper. For example, if the basis weight difference is -12 g / m 2 A basis weight threshold value 126 is set and stored in the memory unit 12.

[0343] The reason why the difference in basis weight between the index for coated paper and the index for plain paper is larger is because the basis weight corresponding to the first transmittance and the basis weight corresponding to the second transmittance are derived using a predetermined judgment criterion (for example, the judgment criterion for plain paper).

[0344] More specifically, the reason why the index for coated paper has a larger difference in basis weight than the index for plain paper is as follows: First, it is assumed that when paper S is plain paper, approximately the same basis weight is derived when either criterion 611 or criterion 711, which are the criteria for plain paper, is used. Also, it is assumed that when paper S is coated paper, approximately the same basis weight is derived when either criterion 612 or criterion 712, which are the criteria for coated paper, is used.

[0345] Next, when the basis weight corresponding to the transmittance of the paper S is derived using the judgment criterion 711 for plain paper, the difference in basis weight is relatively small compared to when the basis weight corresponding to the same transmittance is derived using the judgment criterion 712 for coated paper. An example of a relatively small basis weight difference is the difference between the nominal basis weight of the judgment criterion 711 at a certain transmittance shown in FIG. 57 and the nominal basis weight of the judgment criterion 712 at the same transmittance. Since the difference in basis weight is relatively small, the difference in basis weight corresponding to a certain transmittance is small whether the type of paper S is coated paper or plain paper. In contrast, when the basis weight corresponding to the transmittance of the paper S is derived using the judgment criterion 611 for plain paper, the difference in basis weight between the plain paper and the coated paper is relatively large compared to when the basis weight corresponding to the same transmittance is derived using the judgment criterion 612 for coated paper. An example of a relatively large basis weight difference is the difference between the nominal basis weight of the judgment criterion 611 at a certain transmittance shown in FIG. 56 and the nominal basis weight of the judgment criterion 612 at the same transmittance. The basis weight of coated paper for the same transmittance is greater than that of plain paper, because the transmittance of coated paper tends to be greater than that of plain paper at longer wavelengths.

[0346] According to the above assumption, when the basis weight is derived using the plain paper judgment criterion 611 and the judgment criterion 711, if the paper S is plain paper, the basis weights derived using the respective judgment criteria will be approximately the same. In contrast, if the paper S is coated paper, the basis weight derived using the plain paper judgment criterion 611 will be smaller than the actual basis weight of the paper S (basis weight derived using the coated paper judgment criterion 612). The basis weight derived using the plain paper judgment criterion 711 will be approximately the same as the actual basis weight (basis weight derived using the coated paper judgment criterion 712). Therefore, if the paper S is coated paper, the value obtained by subtracting the basis weight derived using the judgment criterion 711 from the basis weight derived using the judgment criterion 611 will be smaller than the value obtained by subtracting the second basis weight derived using the judgment criterion 711 from the basis weight derived using the judgment criterion 611 when the paper is plain paper.

[0347] The control unit 11 derives the first basis weight and the second basis weight from the first transmittance and the plain paper determination criteria 611 and 711, and determines that the type of paper S is plain paper if the value obtained by subtracting the second basis weight from the first basis weight is equal to or greater than the basis weight threshold 126. Furthermore, the control unit 11 determines that the type of paper S is coated paper if the value obtained by subtracting the second basis weight from the first basis weight is less than the basis weight threshold 126. In other words, since it has been confirmed that the value of the first basis weight minus the second basis weight can be treated as an index for paper type determination and functions advantageously as a determination factor, it has been decided to use it as an input factor for the paper type determination model as described above.

[0348] (Registration profile candidate determination process) FIG. 60 is a diagram showing an example of the paper profile information.

[0349] 60, the paper profile information includes a profile name indicating the name of the registered profile, control parameters including settings of various adjustment values ​​of the image forming apparatus 10 such as paper size and front / back adjustment values, and paper physical property values, which are stored in the storage unit 12 of the image forming apparatus 10 or the storage unit 83 of the server 80 in mutual association. Each item included in the paper physical property values ​​corresponds to each item of the measurement values ​​measured by the media sensor 15.

[0350] The process of determining candidates for registered profiles from the measurement values ​​and the paper profile is performed by comparing the paper physical property values ​​of each registered profile registered in the paper profile with the measurement values ​​acquired by measuring the paper S with the media sensor 15. In other words, the registered profile having the paper physical property values ​​closest to the measurement values ​​is determined as the candidate. An example of a specific determination method will be described below.

[0351] The measured values ​​and the paper physical property values ​​of each registered profile used for the determination are normalized in advance. The distance between the value of each item of the measured values ​​and the value of each item of the paper physical property values ​​of each registered profile is calculated. Here, a weighting coefficient may be set in advance for each item, and the distance may be calculated by multiplying each item by the weighting coefficient. Then, the registered profile with the smallest calculated distance value is determined as a candidate. Note that there may be several variations in basis weight, such as basis weights of 90, 100, 160, 200, 250, and 300, for paper of the same type (brand). Even if the basis weight variations of paper of the same type cannot be determined by the determination using the paper type discrimination model, as described above, by including the basis weight value in the calculation for determining the registered profile, it is possible to select an appropriate registered profile according to the paper type and basis weight of the paper.

[0352] The configuration of the image forming apparatus 10 or the server 80 described above is a main configuration for explaining the features of the above embodiment, and is not limited to the above configuration, and can be modified in various ways within the scope of the claims. Also, configurations that are included in general image forming apparatuses or servers are not excluded. For example, the surface property sensor 60 is disposed on the conveying path 143 of the image forming apparatus 10, but is not limited to this, and may be an apparatus independent of the image forming apparatus 10. In this case, the user can measure the paper characteristics by placing the paper in the irradiation area of ​​the surface property sensor 60. Also, instead of the paper thickness sensor 40, a density sensor capable of measuring the density of the paper may be provided. In this case, the density of the paper is acquired as a measurement value instead of a value related to the paper thickness of the paper.

[0353] 1 and other figures show the image forming apparatus 10 connected to the optional paper feeder 20 and post-processing apparatus 30, but the image forming apparatus 10 may be a standalone apparatus without these options. In addition, in each of the above embodiments, each process described as being executed by the image forming apparatus 10 may be executed by a controller, a PC, or the like connected to the image forming apparatus 10.

[0354] In this embodiment, the paper type determination result is displayed in order of priority ("recommended level") in Fig. 22, but the display of the priority may be omitted. If there is only one candidate (or if the difference between the second and subsequent candidates is a predetermined value or more), in order to improve operability, the selected paper feed tray may be automatically applied, and an operation screen may be displayed that allows the test printing (step S20) operation to be performed immediately (with one touch). [Explanation of symbols]

[0355] 1. Image forming system 10 Image forming device 11 Control section 110 Overall control unit 14 Conveyor section 141, 142 Paper feed tray 143, 144 Transport path 1431 Upper guide a1 aperture 1432 Lower guide 15 Media Sensor 40 Paper thickness sensor 50 Basis weight sensor 60 Surface Sensor 61 Case a2, a3 opening 61a Slope b1, b2, b3 board 62 Light emitting part 63 Collimating Lens 64, 641, 642 Light receiving section 70 Paper pressing section 18 Operation Panel 20 Paper feeder 30 Aftertreatment device 80 Servers 81 Control Unit 82 Memory section 83 Communications Department

Claims

1. A detection unit that detects a value corresponding to a paper characteristic of a recording medium; A learning unit that generates a trained model for paper type discrimination by performing machine learning; A discrimination unit that inputs a value corresponding to the paper characteristic detected by the detection unit into the machine-learned trained model to discriminate the type of the recording medium. the learning unit is provided in a server connected via a network, The detection unit is provided in an image forming apparatus, The server further trains the learning unit to machine-learn the trained model to update the trained model; The discrimination unit is a discrimination system that discriminates the type of the recording medium using the trained model updated on the server.

2. The learning unit generates the trained model by performing machine learning using a value corresponding to the paper characteristic as an input factor of training data and a type of the recording medium as an output factor of the training data. The discrimination system according to claim 1 .

3. The discrimination unit is provided in the image forming apparatus, and discriminates the type of the recording medium using a value detected by the detection unit and the trained model. The discrimination system according to claim 1 or 2.

4. The image forming device acquires the updated trained model, and the discrimination unit discriminates the type of the recording medium using the value detected by the detection unit and the acquired trained model. The discrimination system according to claim 3 .

5. The discrimination unit is provided in a server on a network, and discriminates the type of the recording medium using the trained model on the server. The discrimination system according to claim 1 or 2.

6. The paper characteristics correspond to at least one of a surface property, a thickness or density, and a basis weight of the recording medium. The discrimination system according to any one of claims 1 to 5.

7. The surface property corresponds to a specular reflection light quantity value relating to the amount of light specularly reflected from the surface of the recording medium when the recording medium is irradiated with light from a light source via an optical system at a predetermined incident angle, and a diffuse reflection light quantity value relating to the amount of light diffusely reflected from the surface of the recording medium at at least one reflection angle. The discrimination system according to claim 6.

8. The detection unit is A light source; an optical system that irradiates the light from the light source onto a surface of a recording medium in an irradiation area at a predetermined angle of incidence; a first light receiving unit that detects an amount of light specularly reflected by a surface of the recording medium in the irradiation area; a first detection unit that detects a value related to the surface property of the recording medium, the first detection unit including at least one second light receiving unit that detects an amount of light of the diffuse reflected light that is diffusely reflected at at least one reflection angle on the surface of the recording medium in the irradiation area; a second detection unit that detects a value related to the basis weight of the recording medium and a value related to the thickness or density of the recording medium; The discrimination system according to claim 1 , further comprising:

9. The discrimination system according to claim 1 , wherein the learning unit performs machine learning at a timing when an image forming process is not being performed in the image forming apparatus.

10. The discrimination system according to claim 1 , wherein the control unit of the image forming apparatus does not accept an instruction to execute an image forming process while the learning unit is executing machine learning.

11. an output unit that outputs a discrimination result by the discrimination unit; a reception unit that receives an instruction from a user to change the determination result; A storage unit that stores teacher data for executing the machine learning, The storage unit is provided in the server, The discrimination system according to any one of claims 1 to 10, wherein the learning unit further machine-learns the trained model using change information, including the changed discrimination result and information input into the trained model to obtain the discrimination result, as the teacher data.

12. The discrimination system described in claim 11, wherein the change information transmitted from the image forming device is stored in the memory unit as the training data, and the learning unit further machine-learns the trained model using the stored training data based on predetermined execution conditions.

13. The discrimination system according to claim 1 , wherein the trained model is composed of a random forest.

14. The discrimination system according to claim 1 , wherein the trained model is constituted by a neural network.

15. A method for printing a recording medium comprising the steps of: detecting a value corresponding to a paper characteristic of the recording medium; (b) generating a trained model for paper type discrimination by performing machine learning; and (c) inputting a value corresponding to the paper characteristic detected in the step (a) into the trained model trained by machine learning in the step (b) to discriminate the type of the recording medium, The step (b) is executed by a server connected via a network, The step (a) is performed in an image forming apparatus, In the step (b), the server further performs machine learning on the trained model to update the trained model; A discrimination method, in which in the step (c), the type of the recording medium is discriminated using the trained model updated on the server.

Citation Information

Patent Citations

  • Recording-medium identifying device, recorder, and method for creating recording-medium identifying table

    JP2004050767A

  • Optical sensor and image forming device

    JP2017223692A

  • Image forming apparatus and discrimination device

    JP2018005083A

  • Paper feed parameter management system

    JP2019109316A

  • Image forming apparatus and recording material determination device

    JP2019148673A