Document processing device and document processing system

Abstract

To provide a document processing device which performs more suitable transfer control according to a document placed on a document tray.SOLUTION: The document processing device comprises: the document tray; document transfer means which transfers a document from the document tray; control amount determination means which determines a transfer control amount in transfer of the document, based on classification on the document determined in advance; and estimation means which receives information of the document and estimates the classification on the document. The control amount determination means changes the transfer control amount when a result of the classification estimated by the estimation means and the classification on the document determined in advance are different from each other.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] The present invention relates to a document processing apparatus and a document processing system. [Background technology]

[0002] 2. Description of the Related Art In scanning by an ADF (Automatic Document Feeder) attached to an MFP (Multifunction Peripheral), a technique for setting a paper thickness and controlling the transport of a document in accordance with the paper thickness has been known for some time.

[0003] Patent Document 1 discloses a configuration in which a learning model using machine learning is used to identify the type of document from a scanned image of the document and determine image formation conditions. Summary of the Invention [Problem to be solved by the invention]

[0004] Conventional paper thickness settings require the user to set the paper thickness before starting scanning, and if the user does not know the paper thickness setting, forgets to set the paper thickness, or the thickness of the document placed on the document tray is incorrect, there is a problem that optimal document transport control for each paper thickness setting cannot be achieved. Note that Patent Document 1 does not solve the problem of inability to achieve optimal document transport control for each paper thickness setting.

[0005] An object of the present disclosure is to provide a document processing apparatus that performs more appropriate transport control in accordance with the document placed on the document tray. [Means for solving the problem]

[0006] One embodiment of the present invention is a document processing device comprising a document tray, a document transport means for transporting a document from the document tray, a document information acquisition means for acquiring information about the document, a control amount determination means for determining a transport control amount when transporting the document based on a predetermined classification of the document, and an estimation means for inputting the document information and estimating the classification of the document, wherein the control amount determination means changes the transport control amount when the classification estimated by the estimation means differs from the predetermined classification of the document. [Effects of the Invention]

[0007] According to the present disclosure, it is possible to provide a document processing apparatus that performs more appropriate transport control depending on the document placed on the document tray. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram illustrating an example of a copying machine according to an embodiment. [Figure 2] FIG. 2 is a partial configuration diagram illustrating an example of an enlarged internal configuration of an image forming unit. [Figure 3] FIG. 2 is a partial enlarged view showing an example of a part of a tandem unit made up of four imaging units. [Figure 4] 1 is a perspective view showing an example of a scanner and an ADF of a copying machine according to an embodiment of the present invention. [Figure 5] FIG. 2 is an enlarged view of an example of the main configuration of the ADF together with the upper part of the scanner. [Figure 6] FIG. 2 is a block diagram illustrating an example of a part of an electrical circuit of the copying machine according to the embodiment. [Figure 7] 10 is a flowchart showing an example of a document thickness setting change and a reading operation when the paper thickness is estimated from size information. [Figure 8] FIG. 10 is an image diagram of an example of a document thickness resetting notification screen. [Figure 9] FIG. 1 is a diagram illustrating an example of a step of generating a learning model by machine learning. [Figure 10]FIG. 1 is a diagram illustrating an example of training data to be analyzed by machine learning. [Figure 11] FIG. 10 is a diagram illustrating an example of an inference step using a learning model based on machine learning. [Figure 12] 1 is a flowchart of an example of a specific implementation procedure for completing a learning model by deep learning (DL) using a neural network (NN). [Figure 13] 10A and 10B are diagrams illustrating an example of a threshold value of the degree of coincidence and an estimation result. [Figure 14] FIG. 10 is an image diagram of an example of a document thickness determination level setting screen. [Figure 15] 10 is a flowchart illustrating an example of a screen transition according to the present embodiment. [Figure 16] FIG. 10 is an image diagram of an example of a document thickness setting screen. [Figure 17] FIG. 10 is an image diagram of an example of an automatic document thickness detection setting screen. [Figure 18] 10 is a flowchart illustrating an example of a procedure for changing document thickness settings and a reading operation using a learning model. [Figure 19] FIG. 10 is an explanatory diagram of an example of optimal transport parameters for each paper thickness. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, an embodiment in which the present invention is applied to an electrophotographic copying machine (hereinafter simply referred to as a copying machine) will be described.

[0010] First, the basic configuration of a copying machine according to an embodiment will be described. Fig. 1 is a schematic diagram showing an example of the configuration of a copying machine according to an embodiment. The copying machine shown in Fig. 1 includes an image forming unit 1 as an image forming device, a sheet supply device 40, and an image reading system 50. The image reading system 50 includes a scanner 150 as an image reading device fixed on the image forming unit 1, and an automatic document feeder (hereinafter referred to as ADF) 51 supported by the scanner 150.

[0011] The sheet supply device 40 has two paper feed cassettes 42 arranged in multiple stages in a paper bank 41, a feed roller 43 that feeds recording sheets from the paper feed cassette 42, a separation roller 45 that separates the fed recording sheets one by one, etc. It also has a plurality of conveyance roller pairs 46 that convey the recording sheets to the paper feed path 37 that serves as a conveyance path for the image forming unit 1.

[0012] The sheet cassette 42 contains a stack of recording sheets inside. The topmost sheet of the sheet stack is pressed against the feed roller 43. When the feed roller 43 rotates, the topmost sheet of the sheet stack is fed out of the sheet cassette 42.

[0013] Near the paper feed cassette 42, the first conveying roller of the conveying roller pair 46 and the second conveying roller disposed to the side of it (to the right in the drawing) come into contact with each other to form a conveying nip. In addition, a separation roller 45 is disposed below the first conveying roller and comes into contact with the first conveying roller from below to form a separation conveying nip.

[0014] The recording sheet sent out from the paper feed cassette 42 by the rotational drive of the delivery roller 43 enters a separation conveyance nip formed by the abutment between the first conveyance roller of the conveyance roller pair 46 and a separation roller 45 disposed below the first conveyance roller. In the separation conveyance nip, the first conveyance roller abutting the upper surface of the recording sheet rotates counterclockwise in the drawing, imparting a conveying force to the recording sheet from the paper feed cassette 42 side toward the feed path 44 side. In contrast, the separation roller 45 abutting the lower surface of the recording sheet rotates counterclockwise in the drawing, imparting a conveying force to the recording sheet from the feed path 44 side toward the paper feed cassette 42 side, attempting to return the recording sheet to the paper feed cassette 42.

[0015] When only one recording sheet is fed from the paper feed cassette 42, the first conveying roller and separation roller 45 apply conveying forces in opposite directions to the recording sheet at the separation conveying nip. This causes a load exceeding a predetermined threshold to be applied to the drive transmission system of the separation roller 45. This activates a torque limiter disposed in the drive transmission system, cutting off the transmission of drive force from the DC brushless motor (not shown) to the separation roller 45. This causes the separation roller 45 to rotate along with the recording sheet being conveyed by the first conveying roller, and the recording sheet is discharged from the separation conveying nip toward the feeding path 44.

[0016] On the other hand, when a plurality of recording sheets are fed out from the paper feed cassette 42 in a stacked state, the first conveyance roller imparts a conveying force to the topmost recording sheet in a direction from the paper feed cassette 42 side to the feed path 44 side at the separation conveyance nip. The topmost recording sheet is fed out from the separation conveyance nip towards the feed path 44 side. In response to this, the separation roller 45 imparts a conveying force to the lower recording sheet in a direction from the feed path 44 side to the paper feed cassette 42 side, causing the lower recording sheet to return from the separation conveyance nip towards the paper feed cassette 42 side. As a result, at the separation conveyance nip, the topmost recording sheet is separated from the other recording sheets and fed out to the feed path 44 as a single sheet.

[0017] The recording sheet that has entered the feeding path 44 enters the feeding nip of the pair of feeding rollers 46 and is given a feeding force from the lower side to the upper side in the vertical direction. As a result, the recording sheet is conveyed in the feeding path 44 toward the paper feeding path 37 of the image forming unit 1.

[0018] The image forming unit 1 includes an optical writing device 2 and four imaging units 3K, 3Y, 3M, and 3C that form toner images of black, yellow, magenta, and cyan (K, Y, M, and C). It also includes a transfer unit 24, a paper transport unit 28, a pair of registration rollers 33, a fixing device 34, a switchback device 36, and a paper feed path 37. The optical writing device 2 drives a light source, such as a laser diode or LED (not shown), disposed within the optical writing device 2 to irradiate laser light L onto the four drum-shaped photoconductors 4K, Y, M, and C. The irradiation of the laser light L forms an electrostatic latent image on the surface of the photoconductors 4K, Y, M, and C. This electrostatic latent image is developed into a toner image through a predetermined development process.

[0019] Fig. 2 is an example partial configuration diagram showing an enlarged view of a portion of the internal configuration of the image forming unit 1. Fig. 3 is an example partial enlarged view showing a portion of a tandem unit consisting of four imaging units 3K, Y, M, and C. Note that the four imaging units 3K, Y, M, and C have almost the same configuration except for the toner colors they use, so the subscripts K, Y, M, and C attached to each reference number are omitted in Fig. 3.

[0020] The imaging units 3K, Y, M, and C each have a photoconductor 4K, Y, M, or C and various devices disposed around the photoconductor 4K supported on a common support as a single unit, and are detachable from the main body of the image forming unit 1. Taking the black imaging unit 3K as an example, the imaging unit 3K has a charging device 5, a developing device 6, a drum cleaning device 15, and a static elimination lamp 22, etc., around the photoconductor 4K. In this copier, the four imaging units 3K, Y, M, and C are arranged facing each other along the endless movement direction of the intermediate transfer belt 25, which will be described later, in a so-called tandem configuration.

[0021] The photoreceptors 4K, Y, M, and C are drum-shaped, with a photosensitive layer formed by coating an organic photosensitive material on a bare tube made of aluminum or the like, although endless belt-shaped ones may also be used.

[0022] The developing devices 6K, Y, M, and C are configured to develop latent images using a two-component developer containing a magnetic carrier and non-magnetic toner (not shown). They have an agitation unit 7 that agitates and transports the two-component developer contained therein and supplies it to the developing sleeve 12, and a developing unit 11 that transfers the toner in the two-component developer carried on the developing sleeve 12 to the photoconductors 4K, Y, M, and C.

[0023] The stirring section 7 is located at a lower position than the developing section 11, and includes two transport screws 8 arranged parallel to each other, a partition plate provided between the transport screws 8, and a toner concentration sensor 10 provided on the bottom surface of the developing case 9.

[0024] The developing unit 11 includes a developing sleeve 12 that faces the photoconductors 4K, Y, M, and C through the opening of the developing case 9, a magnet roller 13 that is non-rotatably provided inside the developing sleeve 12, and a doctor blade 14 that brings its tip close to the developing sleeve 12. The developing sleeve 12 is a non-magnetic, rotatable cylinder.

[0025] The magnet roller 13 has a plurality of magnetic poles that are arranged in sequence from the position facing the doctor blade 14 toward the rotation direction of the developing sleeve 12. These magnetic poles each apply a magnetic force to the two-component developer on the developing sleeve 12 at a predetermined position in the rotation direction. This attracts and carries the two-component developer sent from the agitator 7 to the surface of the developing sleeve 12, and forms a magnetic brush on the surface of the developing sleeve 12 along the magnetic lines of force.

[0026] The magnetic brush is regulated to an appropriate layer thickness when it passes a position facing the doctor blade 14 as the developing sleeve 12 rotates, and is then transported to the developing area facing the photoconductors 4K, Y, M, and C. The toner is then transferred onto the electrostatic latent image due to the potential difference between the developing bias applied to the developing sleeve 12 and the electrostatic latent image on the photoconductors 4K, Y, M, and C, contributing to development.

[0027] Furthermore, as the developing sleeve 12 rotates, the two-component developer returns to the developing unit 11, and after being separated from the surface of the sleeve by the influence of the repulsive magnetic field formed between the magnetic poles of the magnet roller 13, it is returned to the agitating unit 7. In the agitating unit 7, an appropriate amount of toner is replenished to the two-component developer based on the detection result of the toner concentration sensor 10. Note that the developing device 6 may be one that uses a one-component developer that does not contain a magnetic carrier, instead of one that uses a two-component developer.

[0028] The drum cleaning devices 15K, Y, M, and C use a type in which a cleaning blade 16 made of an elastic material is pressed against the photoreceptor 4, but other types may also be used. In order to improve cleaning performance, this example uses a type in which a contact-conductive fur brush 17, whose outer circumferential surface contacts the photoreceptor 4K, Y, M, and C, is rotatable in the direction of the arrow in the figure.

[0029] The fur brush 17 also serves to scrape lubricant from a solid lubricant (not shown) and finely powder it before applying it to the surfaces of the photoreceptors 4K, Y, M, and C. A metal electric field roller 18, which applies a bias to the fur brush 17, is rotatable in the direction indicated by the arrow in the figure. The tip of a scraper 19 is pressed against the electric field roller 18. Toner adhering to the fur brush 17 is transferred to the electric field roller 18, which rotates in a counter-direction relative to the fur brush 17 and to which a bias is applied. After being scraped off the electric field roller 18 by the scraper 19, the recovered toner falls onto a recovery screw 20. The recovery screw 20 transports the recovered toner toward the end of the drum cleaning device 15, perpendicular to the plane of the drawing, and transfers it to an external recycling transport device. The recycling transport device then sends the transferred toner to the developing devices 6K, Y, M, and C for recycling.

[0030] The static elimination lamp 22 eliminates static electricity from the photoconductors 4K, Y, M, and C by irradiating them with light. The surfaces of the photoconductors 4K, Y, M, and C that have been neutralized are then uniformly charged by the charging device 5, and then optical writing processing is performed by the optical writing device 2. The charging device 5 shown in FIG. 3 uses a charging roller to which a charging bias is applied that rotates while being in contact with the photoconductors 4K, Y, M, and C. The charging device 5 may also be a scorotron charger or the like that charges the photoconductors 4K, Y, M, and C without contact.

[0031] K, Y, M, and C toner images are formed on the photoconductors 4K, Y, M, and C of the imaging units 3K, Y, M, and C by the process described above. A transfer unit 24 is disposed below the imaging units 3K, Y, M, and C. The transfer unit 24 moves an intermediate transfer belt 25, which is stretched by multiple rollers, endlessly in a clockwise direction in the drawing while contacting the photoconductors 4K, Y, M, and C. This forms primary transfer nips for K, Y, M, and C where the photoconductors 4K, Y, M, and C contact the endless intermediate transfer belt 25.

[0032] Near the primary transfer nips for K, Y, M, and C, primary transfer rollers 26K, Y, M, and C arranged inside the belt loop press the intermediate transfer belt 25 toward the photoconductors 4K, Y, M, and C. A primary transfer bias is applied to each of these primary transfer rollers 26K, Y, M, and C by a power supply (not shown). As a result, a primary transfer electric field is formed in the primary transfer nips for K, Y, M, and C, which electrostatically moves the toner images on the photoconductors 4K, Y, M, and C toward the intermediate transfer belt 25.

[0033] As the intermediate transfer belt 25 moves endlessly clockwise in the drawing, it passes through the primary transfer nips for K, Y, M, and C in sequence, and toner images are sequentially superimposed and primarily transferred at each primary transfer nip onto the front surface of the intermediate transfer belt 25. As a result of this superimposed primary transfer, a four-color superimposed toner image (hereinafter referred to as a four-color toner image) is formed on the front surface of the intermediate transfer belt 25.

[0034] Below the transfer unit 24 in the figure, there is provided a paper transport unit 28 that loops an endless paper transport belt 29 between a drive roller 30 and a secondary transfer roller 31 and moves it endlessly. The intermediate transfer belt 25 and the paper transport belt 29 are sandwiched between the secondary transfer roller 31 itself and a lower tension roller 27 of the transfer unit 24. This forms a secondary transfer nip where the front surfaces of the intermediate transfer belt 25 and the paper transport belt 29 come into contact with each other. A secondary transfer bias is applied to the secondary transfer roller 31 by a power source (not shown). Meanwhile, the lower tension roller 27 of the transfer unit 24 is grounded. This forms a secondary transfer electric field in the secondary transfer nip.

[0035] A pair of registration rollers 33 is disposed on the right side of the secondary transfer nip in the drawing. A registration roller sensor (not shown) is disposed near the entrance of the registration nip of the pair of registration rollers 33. When a recording sheet is conveyed from the sheet supply device 40 toward the pair of registration rollers 33, the conveyance of the recording sheet is temporarily stopped after a predetermined time has passed since the leading edge of the recording sheet was detected by the registration roller sensor, and the leading edge of the recording sheet abuts against the registration nip of the pair of registration rollers 33.

[0036] When the leading edge of the recording sheet strikes the registration nip, the pair of registration rollers 33 resume roller rotation at a timing that allows the recording sheet to be synchronized with the four-color toner image on the intermediate transfer belt 25, and sends the recording sheet to the secondary transfer nip. Within the secondary transfer nip, the four-color toner image on the intermediate transfer belt 25 is secondarily transferred all at once onto the recording sheet by the action of the secondary transfer electric field or nip pressure, and combined with the white color of the recording sheet, forms a full-color image. After passing through the secondary transfer nip, the recording sheet separates from the intermediate transfer belt 25 and is conveyed to the fixing device 34 as the paper transport belt 29 moves endlessly while being held on the front surface of the paper transport belt 29.

[0037] After passing through the secondary transfer nip, residual toner that was not transferred to the recording sheet at the secondary transfer nip adheres to the surface of the intermediate transfer belt 25. The residual toner is scraped off and removed by a belt cleaning device 32 that contacts the intermediate transfer belt 25.

[0038] The recording sheet conveyed to the fixing device 34 has the full-color image fixed thereon by pressure and heat, and is then sent from the fixing device 34 to a pair of paper discharge rollers 35 and then discharged outside the machine.

[0039] 1, a switchback device 36 is disposed below the paper transport unit 28 and the fixing device 34. As a result, after the image fixing process on one side of the recording sheet has been completed, the path of the recording sheet is switched by a switching claw to the switchback device 36 side, where it is reversed and enters the secondary transfer nip again. Then, the recording sheet undergoes the secondary image transfer process and fixing process on the other side, and is then discharged onto the paper output tray.

[0040] Scanner 150, which is fixed on image forming unit 1, has a movable reading unit 152. Scanner 150 and ADF 51 each have a fixed reading unit. Mobile reading unit 152 is disposed directly below second contact glass 154 (see FIG. 4), which is fixed to the upper wall of the casing of scanner 150 so as to come into contact with document MS, and can move an optical system consisting of a light source, a reflecting mirror, etc., left and right in the drawing. As the optical system moves from left to right in the drawing, light emitted from the light source is reflected by document MS placed on second contact glass 154, passes through multiple reflecting mirrors, and is then received by image reading sensor 153, which is fixed to the scanner body.

[0041] Meanwhile, the fixed reading unit is comprised of a first-side fixed reading unit 151 disposed inside the scanner 150 and a second-side fixed reading unit 152 disposed inside the ADF 51. The first-side fixed reading unit 151, which has a light source, a reflecting mirror, an image reading sensor such as a CCD, and the like, is disposed directly below a first contact glass 155 (see FIG. 4) fixed to the upper wall of the casing of the scanner 150 so as to come into contact with the original MS. When a sheet-like original MS transported by the ADF 51 (described later) passes over the first contact glass 155, light emitted from the light source is sequentially reflected by the original surface, passes through multiple reflecting mirrors, and is received by the image reading sensor. This allows the first side of the original MS to be scanned without moving the optical system consisting of the light source, reflecting mirrors, and the like. The second-side fixed reading unit 152 scans the second side of the original MS after it has passed the first-side fixed reading unit 151.

[0042] The ADF 51 disposed above the scanner 150 has a document placement table 53, which is a document placement means for placing a document MS before reading, in a main body cover 52. The ADF 51 also has a transport unit 54 for transporting the document MS as a sheet member, a document stack table 55 for stacking the document MS after reading, and the like.

[0043] FIG. 4 is a perspective view of an example of a scanner and ADF of a copying machine according to this embodiment. As shown in FIG. 4, the ADF 51 is supported by hinges 159 fixed to the scanner 150 so as to be able to swing up and down. The ADF 51 swings like a door, exposing a first contact glass 155 and a second contact glass 154 on the top surface of the scanner 150 when open. In the case of single-sided bound documents, such as a book bound at one corner of a stack of documents, the documents cannot be separated one by one and therefore cannot be transported by the ADF. Therefore, in the case of single-sided bound documents, the ADF 51 is opened as shown in FIG. 4, and the single-sided bound document with the page to be read open is placed face down on the second contact glass 154, and the ADF is then closed. The image of the page is then read by the moving reading unit 152 of the scanner 150, as shown in FIG. 1.

[0044] On the other hand, in the case of a stack of documents consisting of multiple independent documents MS stacked on top of each other, the documents MS can be automatically transported one by one by the ADF 51, and sequentially read by the first surface fixed reading unit 151 in the scanner 150 and the second surface fixed reading unit 152 in the ADF 51.

[0045] In this case, after placing a stack of documents on the document placing table 53, the copy start key 158 is pressed. Then, the ADF 51 sends the documents MS of the document stack placed on the document placing table 53 into the transport unit 54 in order from the top, and transports them toward the document stack table 55 while inverting them. During this transport process, immediately after inverting the documents MS, they are passed directly above the first-side fixed reading unit 151 of the scanner 150. At this time, the image of the first side of the documents MS is read by the first-side fixed reading unit 151 of the scanner 150.

[0046] 5 is an enlarged view of an example of the main configuration of the ADF 51, along with the upper part of the scanner 150. The ADF 51 includes a document setting unit A, a separation and feeding unit B, a registration unit C, a turning unit D, a first reading and conveying unit E, a second reading and conveying unit F, a paper discharge unit G, and a stacking unit H. The ADF 51 also includes a document conveying path for conveying the document MS from the document placing table 53 toward the first-side fixed reading unit 151, which is the image reading position.

[0047] The document setting unit A includes a document placement tray 53 on which a stack of documents MS is set. The separation and feeding unit B separates and feeds documents MS one by one from the stack of documents MS set therein. The registration unit C temporarily bumps against the fed documents MS, aligns them, and then sends them out. The turning unit D has a curved conveying section that curves in a C shape, and turns the documents MS upside down while folding them back within this curved conveying section. The first reading and conveying unit E conveys the documents MS on the first contact glass 155 and causes the first side of the documents MS to be read by the first-side fixed reading unit 151, which is disposed inside the scanner below the first contact glass 155. The second reading and conveying unit F conveys the documents MS below the second fixed reading unit 95 and causes the second side of the documents MS to be read by the second fixed reading unit 95. The paper discharge unit G discharges the documents MS, whose images on both sides have been read, toward the stack unit H. The stacking section H is for stacking the documents MS on a document stacking table 55 .

[0048] The originals MS are set with their leading edge placed on the movable original table 60, which can swing in the directions of arrows a and b in the figure according to the thickness of the stack of originals MS, and their trailing edge placed on the original placement table 53. The originals MS are positioned on the original placement table 53 in the width direction (the direction perpendicular to the plane of the drawing) by abutting side guides (not shown) at both ends of the originals MS. The originals MS set in this manner push up a lever member 62, which is swingably disposed above the movable original table 60. Accordingly, the original placement sensor 63 detects the placement of the originals MS and transmits a detection signal to the controller 904 (see FIG. 6). The detection signal is then sent from the controller 904 to the scanner's reading control unit 903 via the I / F.

[0049] The document placing platform 53 holds a first length sensor 57, a second length sensor 58, a third length sensor 202, and a fourth length sensor 201, each consisting of a reflective photosensor or an actuator-type sensor that detects the length of the document MS in the transport direction. The third length sensor 202 is, for example, a check length sensor. The fourth length sensor 202 is, for example, a business card length sensor. The following describes an example in which the third length sensor 202 is a check length sensor and the fourth length sensor 202 is a business card length sensor.

[0050] The length of the document MS in the transport direction is detected by first length sensor 57, second length sensor 58, third length sensor 202, and fourth length sensor 201. Third length sensor 202 is positioned so that it does not turn ON when a check is placed on the document tray. Fourth length sensor 201 is positioned so that it does not turn ON when a business card is placed on the document tray.

[0051] It is possible to detect whether a document of a specific size, such as a business card or check, has been placed based on the detection information from first length sensor 57, second length sensor 58, third length sensor 202, and fourth length sensor 201. As a guide for the positions of third length sensor 202 and fourth length sensor 201, since a check is 185 mm long and a business card is 91 mm long, third length sensor 202 should be positioned approximately 190 mm and fourth length sensor 201 should be positioned 96 mm from the fence that the leading edge of the document hits when it is placed on the document tray.

[0052] A pickup roller 80 is disposed above the stack of documents MS placed on the movable document table 60 and is supported by a cam mechanism so as to be movable in the vertical direction (the directions of arrows c and d in the figure). This cam mechanism is driven by a pickup motor 56, thereby enabling the pickup roller 80 to move up and down. When the pickup roller 80 moves upward, the movable document table 60 also swings in the direction of arrow a in the figure, and the pickup roller 80 comes into contact with the topmost document MS in the stack of documents MS.

[0053] As the movable original table 60 rises further, the table rise detection sensor 59 eventually detects that the movable original table 60 has reached its upper limit, which stops the pickup motor 56 and stops the upward movement of the movable original table 60.

[0054] The user (operator) operates an operation unit 902, which is provided on the main body of the copier and is composed of a numeric keypad 160, a display 161, etc., to set the reading mode, indicating whether the reading mode is double-sided or single-sided, and presses the copy start key 158. When the user presses the copy start key 158, the controller 904 of the ADF 51 receives a document feed signal from the main body control unit 901 (see FIG. 6). Then, the pickup roller 80 is rotated by the forward rotation of the paper feed motor 76, and the document MS on the movable document table 60 is fed out from the movable document table 60.

[0055] When setting the double-sided reading mode or the single-sided reading mode, it is possible to collectively set the double-sided reading mode or the single-sided reading mode for all of the originals MS placed on the movable original table 60. It is also possible to set the reading mode individually for each of the originals MS, for example, by setting the first and tenth originals MS to the double-sided reading mode and the other originals MS to the single-sided reading mode.

[0056] The document MS sent by the pickup roller 80 enters the separation and conveyance section B and is sent to a contact position with the paper feed belt 84. The paper feed belt 84 is stretched by a drive roller 82 and the like, and is moved endlessly in the clockwise direction in the figure by the rotation of the drive roller 82 accompanying the forward rotation of the paper feed motor 76. A separation roller 85, which is driven to rotate clockwise in the figure by the forward rotation of the paper feed motor 76, contacts the lower stretching surface of the paper feed belt 84. At the contact point, the surface of the paper feed belt 84 moves in the paper feed direction. In contrast, the separation roller 85 contacts the paper feed belt 84 with a predetermined pressure. When the separation roller 85 is in direct contact with the paper feed belt 84, or when only one document MS is sandwiched in the contact area, the separation roller 85 rotates along with the belt or the document MS. However, when multiple sheets of originals MS are sandwiched in the contact portion, the accompanying rotation force becomes lower than the torque of the torque limiter, and the rollers are rotated in the clockwise direction in the figure, which is opposite to the accompanying rotation direction. As a result, the separation roller 85 applies a moving force in the opposite direction to the paper feeding to the sheets of originals MS below the top, and only the topmost sheet of originals MS is separated from the several sheets of originals (hereinafter, the operation up to this point is referred to as the paper feeding / separation operation).

[0057] The document MS, separated into a single sheet by the action of the paper feed belt 84 and separation roller 85, enters the registration section C. Then, as the document MS passes directly below the abutment sensor 72, its leading edge is detected by the abutment sensor 72. At this time, the pickup roller 80, which receives the driving force of the pickup motor 56, is still rotating. However, the pickup roller 80 is separated from the document MS by the downward movement of the movable document table 60, and the document MS is transported solely by the endless moving force of the paper feed belt 84. The endless movement of the paper feed belt 84 continues for a predetermined time from the timing when the leading edge of the document MS is detected by the abutment sensor 72. As a result, the leading edge of the document MS abuts against the contact portion between the pull-out drive roller 86 and the pull-out driven roller 87, which is driven to rotate while abutting against the pull-out drive roller 86. At this time, by separating the document MS from the document MS based on the timing when the leading edge of the document MS is detected by the abutment sensor 72, the timing when the document MS starts to be transported solely by the endless moving force of the paper feed belt 84 can be advanced or delayed. By advancing or delaying the timing at which the document MS begins to be transported solely by the endless moving force of the paper feed belt 84, it is possible to adjust the amount of contact at which the document MS strikes the contact point with the pull-out driven roller 87.

[0058] If the document speed at this time is 500 mm / s, then in order to increase the abutment amount by 1 mm, it is possible to delay the timing of separation from the document MS by 2 ms (500 mm / s x 0.002 s = 1 mm). The pull-out driven roller 87 serves to transport the document MS to the pair of intermediate rollers 66 downstream in the document transport direction, and is rotationally driven by the reverse rotation of the paper feed motor 76. When the paper feed motor 76 rotates reversely, the pull-out driven roller 87 and one of the rollers of the pair of intermediate rollers 66 that are in contact with each other begin to rotate, and the endless movement of the paper feed belt 84 stops. At this time, the rotation of the pickup roller 80 also stops.

[0059] The document MS sent out from the pull-out driven roller 87 passes directly below the document width sensor 73. The document width sensor 73 has multiple paper detection units made up of reflective photosensors or the like. These paper detection units are aligned in the document width direction (the direction perpendicular to the plane of the drawing). The width size of the document MS is detected based on which paper detection unit detects the document MS. The length of the document MS in the transport direction is detected based on the timing from when the leading edge of the document MS is detected by the abutment sensor 72 to when the trailing edge of the document MS is no longer detected by the abutment sensor 72.

[0060] The leading edge of the document MS, whose widthwise size has been detected by the document width sensor 73, enters the turning section D and is sandwiched in the contact area between the rollers of the intermediate roller pair 66. The transport speed of the document MS by this intermediate roller pair 66 is set to be faster than the transport speed of the document MS in the first reading and transporting section E, which will be described later. This shortens the time it takes to send the document MS to the first reading and transporting section E.

[0061] In this way, the size of the original MS can be detected by the original width sensor 73, the first length sensor 57, the second length sensor 58, the third length sensor 202, and the fourth length sensor 201. On the other hand, the user can also specify the size (read size) of the original MS using the operation unit 902. For example, the user can select and specify the size of a business card (55 mm x 91 mm) or a check (85 mm x 185 mm) from a variety of size options.

[0062] The leading edge of the document MS being transported within the turning section D passes a position facing the reading entrance sensor 67. When the leading edge of the document MS is detected by the reading entrance sensor 67, the document transport speed by the intermediate roller pair 66 is reduced until the leading edge reaches the position of the reading entrance roller pair (the pair of 89 and 90) on the downstream side in the transport direction. In addition, as the reading motor 77 (see FIG. 6) starts to be rotated, one roller of the reading entrance roller pair, one roller of the reading exit roller pair 92, and one roller of the second reading exit roller pair 93 each start to be rotated.

[0063] Within the turning section D, the document MS is reversed upside down and the transport direction is reversed while being transported on the curved transport path between the pair of intermediate rollers 66 and the pair of reading entrance rollers. Then, the leading edge of the document MS that has passed through the nip between the pair of reading entrance rollers passes directly below the registration sensor 65. Hereinafter, the operation up to this point after the paper feeding / separation operation is referred to as the pull-out operation.

[0064] When the leading edge of the document MS is detected by the registration sensor 65, the document transport speed is decelerated over a predetermined transport distance. Then, the transport motor 192 (see FIG. 6) is stopped to stop the rotation of the pull-out drive roller 86 and the pair of intermediate rollers 66, and the reading motor 77 is stopped to stop the rotation of the pair of reading entrance rollers. As a result, the transport of the document MS is temporarily stopped at the registration position in front of the first reading transport unit E. In addition, a registration stop signal is sent to the reading control unit 903.

[0065] 6 is a block diagram showing an example of a portion of the electrical circuitry of a copying machine according to an embodiment. A main body control unit 901 of the image forming unit 1 and a reading control unit 903 of the scanner each include a CPU, RAM, ROM, etc. A controller 904 of the ADF 51 also includes a CPU, RAM, ROM, etc. The main body control unit 901 and the reading control unit 903, the reading control unit 903 and the controller 904, and the controller 904 and the main body control unit 901 are all connected to each other and are capable of communicating with each other.

[0066] A transport motor 192 connected to the controller 904 is a rotational drive source for the pull-out drive roller 86 and the pair of paper discharge rollers 94 in the ADF 51. A pull-out clutch 193 connected to the controller 904 connects or disconnects the rotational drive force of the transport motor 192 to or from the pull-out drive roller 86. A paper discharge clutch 194 connects or disconnects the rotational drive force of the transport motor 192 to or from the pair of paper discharge rollers 94, which are the feed-out transport means.

[0067] Upon receiving the registration stop signal from the controller 904, the reading control unit 903 transmits a reading start signal, which is a paper feed permission signal, to the controller 904. Upon receiving the reading start signal, which is a paper feed permission signal, the controller 904 resumes rotation of the transport motor 192 and the reading motor 77. Then, at the timing when the leading edge of the original MS, calculated based on the pulse count of the reading motor 77, reaches the reading position by the first-side fixed reading unit 151, the controller 904 transmits a gate signal, which indicates the sub-scanning direction effective image area of ​​the first side of the original MS, to the reading control unit 903. This transmission continues until the trailing edge of the original MS leaves the reading position by the first-side fixed reading unit 151, and the first side of the original MS is read by the first-side fixed reading unit 151.

[0068] After passing through the first reading conveyance unit E, the document MS passes through the reading exit roller pair 92, and then its leading edge is detected by the paper discharge sensor 61. When the single-sided reading mode is set, there is no need to read the second side of the document MS by the second fixed reading unit 95. Therefore, when the leading edge of the document MS is detected by the paper discharge sensor 61, the driving force of the conveyance motor 192 is connected to the paper discharge roller pair 94 by the paper discharge clutch 194. Then, based on the pulse count of the paper discharge motor after the leading edge of the document MS is detected by the paper discharge sensor 61, the timing at which the trailing edge of the document MS will leave the nip of the paper discharge roller pair 94 is calculated. Based on the result of this calculation, the operation of the paper discharge clutch 194 is stopped.

[0069] On the other hand, when the double-sided reading mode is set, the timing from when the leading edge of the original MS is detected by the paper discharge sensor 61 until the original reaches the second fixed reading unit 95 is calculated based on the pulse count of the reading motor 77. Then, the controller 904 transmits a gate signal indicating the effective image area in the sub-scanning direction on the second side of the original MS to the reading control unit 903 at the timing of the calculation result. This transmission continues until the trailing edge of the original MS leaves the reading position by the second fixed reading unit 95, and the second side of the original MS is read by the second fixed reading unit 95.

[0070] The second fixed reading unit 95 is composed of a contact image sensor (CIS), and its reading surface is coated to prevent vertical reading streaks caused by glue-like foreign matter adhering to the document MS adhering to the reading surface. A second reading roller 96 is disposed opposite the second fixed reading unit 95 as a document support means for supporting the document MS from the non-reading side. The second reading roller 96 prevents the document MS from floating at the reading position by the second fixed reading unit 95, and also functions as a reference white area for acquiring shading data in the second fixed reading unit 95.

[0071] FIG. 7 is a flowchart showing an example of a document thickness setting change and a reading operation when the paper thickness is estimated from size information.

[0072] In step S1, the user places the document to be copied or scanned on the document tray and presses the copy start key 158 or scan start button. The ADF 51, which is an example of a document processing device, starts the first step of processing. The first step is a paper feeding and reading operation that includes processing to estimate the paper thickness or paper type.

[0073] In step S2, the ADF 51 starts feeding the document. The controller 904 included in the ADF 51 determines the size of the document using the detection information from the document width sensor 73, the first length sensor 57, the second length sensor 58, the third length sensor 202, and the fourth length sensor 201. The controller 904 also acquires from the image forming unit 1 the document size (user setting) specified by the user via the operation unit 902.

[0074] In step S3, the controller 904 estimates the paper thickness or paper type from the determined or acquired document size. The basis for the estimation in step S3 is that for certain types of documents, such as business cards, checks, or receipts, the document size, paper thickness (business cards are thick paper, checks and receipts are thin paper, etc.), and paper type are fixed, and past experience has shown that the document size can be information that can be used to determine the paper thickness or paper type with a high degree of probability.

[0075] In this embodiment, business cards (≒ thick paper) and checks (≒ thin paper) are handled, but if there is another document whose paper thickness or paper type can be determined from its size, it is possible to identify the paper thickness and paper type in the same way as business cards and checks. An example of identifying paper thickness is described below.

[0076] In step S4, the controller 904 branches the process depending on whether the paper thickness can be estimated or not, since depending on the size of the document, it is not always possible to estimate the paper thickness from size information, such as for business cards or checks.

[0077] If the estimation is not possible, the controller 904 does not change the currently set document thickness information, and proceeds to the paper feed and reading operation in the second step of step S10. Note that the second step is a paper feed and reading operation that does not include processing for estimating the paper thickness or paper type. If the estimation is possible, the controller 904 proceeds to step S5 and compares the estimated paper thickness with the paper thickness currently set in the document thickness information.

[0078] If the estimated paper thickness is the same as the paper thickness currently set in the document thickness information, there is no need to reset the paper thickness currently set in the document thickness information, so the document thickness information is not changed and the paper feeding and reading operation proceeds to the second step of step S10.

[0079] If the estimated paper thickness differs from the paper thickness currently set in the document thickness information, the controller 904 determines that there is a high possibility that the user has set the paper thickness in the document thickness information incorrectly, and proceeds to step S6.

[0080] In step S6, the controller 904 determines whether automatic resetting of the paper thickness is permitted (instructed). If automatic resetting of the paper thickness is permitted, the controller 904 proceeds to step S9, where it resets (changes) the paper thickness set in the document thickness information. In step S10, the controller 904 starts paper feeding and reading operations with a transport control amount that is optimal for the paper thickness set in the changed document thickness information.

[0081] If automatic resetting of paper thickness is not permitted, the controller 904 proceeds to step S7 and suspends the paper feeding and reading operations. The controller 904 displays, for example, an original thickness reset notification screen 1000 shown in FIG. 8 on the display unit 905, prompting the user to confirm and input whether or not to reset (change) the paper thickness set in the original thickness information. The original thickness reset notification screen 1000 of FIG. 8 is displayed when the degree of match, described below, exceeds a threshold and the original thickness setting is determined to be incorrect. The original thickness reset notification screen 1000 of FIG. 8 makes the user aware of an incorrect or forgotten original thickness setting, enabling the user to set the correct original thickness. The document thickness reset notification screen 1000 in FIG. 8 will be described in detail later.

[0082] If the user does not reset the paper thickness set in the document thickness information, the controller 904 proceeds to the second step of step S10 for paper feeding and reading operations. If the user resets the paper thickness set in the document thickness information, the controller 904 proceeds to step S9 and changes the paper thickness set in the document thickness information. In step S10, the controller 904 starts paper feeding and reading operations with a transport control amount that is optimal for the paper thickness set in the changed document thickness information.

[0083] The following describes steps for generating a learning model 2000 by machine learning in this embodiment. Fig. 9 is a diagram for explaining an example of steps for generating a learning model 2000 by machine learning.

[0084] In this embodiment, a learning model 2000 generated by machine learning is implemented in a control program written in the controller 904 of the ADF 51, which is an example of a document processing apparatus. The learning model 2000 is generated by analyzing training data 2010 created from collected data using a cloud server 908 or an external PC (see FIG. 6) capable of generating the learning model 2000. The generated learning model 2000 is a type of calculation algorithm and is modularized and implemented as part of the control program. While FIG. 9 illustrates a case in which the learning model 2000 is implemented in the controller 904 of the ADF 51, it may also be implemented in the main body control unit 901 on the image forming unit 1 side, or in the cloud server 908 that can communicate via a network using the external communication I / F 907 (see FIG. 6). A method for using the implemented learning model 2000 will be described later. The controller 904 that executes the control program functions as a control amount determination means and an estimation means. The controller 904 that executes the control program may also function as a notification means, an input means, and a document information acquisition means.

[0085] FIG. 10 is a diagram illustrating an example of training data 2010 analyzed by machine learning. In this embodiment, image data type and size information, which are document information to be read, are used as input data, and paper thickness, which is the correct answer information to be predicted in the inference step, is used as output data for analysis by machine learning. In FIG. 10, paper thickness (thick paper, plain paper, and thin paper) is shown as multiple classifications of documents. In addition to paper thickness, paper type (plain paper, coated paper, and carbon paper) can also be considered as multiple classifications of documents. Here, an example will be described in which multiple classifications of documents are paper thickness.

[0086] In the training data 2010 in Figure 10, the input data and output data were set with a focus on paper thickness, but if the image data type and size information for each paper type are prepared as manuscript information in a similar manner, analysis can be performed using machine learning.

[0087] In addition to the image data type, size information (user setting) and size information (sensor information) are added to the document information because, for example, business cards, which are often made of thick paper, and checks or receipts, which are often made of thin paper, have fixed sizes, and size information can be used to more accurately determine paper thickness than image data type.

[0088] An inference step using the learning model 2000 based on machine learning in this embodiment will be described. FIG. 11 is a diagram illustrating an example of the inference step using the learning model 2000 based on machine learning. In this embodiment, when an original is actually read, the paper thickness is predicted using the learning model 2000 generated by machine learning. In the inference step, the original information (size information) set in the image forming unit 1 and the read image data are input as input data to the learning model 2000, and the degree of match (similarity) with each paper thickness (thick paper, regular paper, thin paper) is received as output data.

[0089] The control program written in the controller 904 has preset therein the optimum paper feed and transport control amounts for each thickness of the original so that skew and paper feed failures are unlikely to occur.

[0090] The controller 904 that executes the control program determines whether the degree of coincidence with each paper thickness output from the learning model 2000 matches the paper thickness set by the user. If the paper thickness inferred from the degree of coincidence with each paper thickness output from the learning model 2000 differs from the paper thickness set by the user, the controller 904 automatically changes the setting to a paper thickness that matches more closely, or notifies the user to prompt them to change the paper thickness setting, and changes the amount of document transport control.

[0091] FIG. 12 is a flowchart of an example of a specific implementation procedure for completing a learning model 2000 by deep learning (DL) using a neural network (NN).

[0092] In step S31, the designer collects training data 2010. Specifically, the designer uses the ADF 51 to read a document, and stores the scanned image data, the classification (paper thickness) of the document used, size information (user setting), and size information (sensor information) as a set. The classification (paper thickness) of the document used becomes the correct answer information to be inferred, and becomes labeling data.

[0093] It is desirable to use data that is likely to be used by the user in practice as the training data 2010. Therefore, it is most desirable if the user can provide the classification and size information of the document that has been read in an actual workplace such as an office, as well as the scanned image data.

[0094] If this is not possible, it may be acceptable to prepare a large amount of image data collected by the designer on the web and data on the classification of manuscripts (paper thickness) that is assumed to correspond to it based on experience (for example, a dataset of image data of business card contents + "thick paper" information, or a dataset of image data of check contents + "thin paper" information). In deep learning, the more training data 2010 there is, the better, so it is advisable to collect as much data as possible in step S1.

[0095] In step S32, the designer generates a learning model 2000 by deep learning using a neural network, using the training data 2010 collected in step S31. The learning model 2000 can be generated by using a general AI framework. For example, TensorFlow (registered trademark), MATLAB (registered trademark), PyTorch (registered trademark), or ONNX (registered trademark) can be used as the AI ​​framework.

[0096] The learning model 2000 generated by the AI ​​framework is often not in a format that can be processed by the CPU of the ADF 51 or the image forming unit 1 as is, so in step S33 it is converted into embedded code (mainly C language, etc.) and written to a board incorporating a CPU.

[0097] Conversion to embedded code can be done using conversion tools provided by CPU vendors. If you use a proprietary CPU, you may need to use a proprietary method to convert to embedded code.

[0098] When the inference step using the learning model 2000 is performed on a server on the cloud, such as the cloud server 908, rather than on an embedded CPU, the conversion and writing to embedded code in step S33 is not necessary. When the inference step using the learning model 2000 is performed on a server on the cloud, such as the cloud server 908, rather than on an embedded CPU, it may be necessary to convert the model to a format that can be executed on the cloud server 908 instead. If conversion to a format that can be executed on the cloud server 908 is necessary, this is performed.

[0099] In step S34, with the converted learning model 2000 incorporated, the controller 904 of the ADF 51 or the main body control unit 901 of the image forming unit 1 performs an actual inference step, outputs the degree of match for each paper thickness, and records the processing time actually required. The document to be read in step S34 must be unknown to the learning model 2000, not the document read in step S31.

[0100] In step S34, the designer collects a certain number of (hundreds or thousands of) pieces of data on the degree of match and processing speed. In step S35, the designer analyzes the collected data and adjusts the threshold value for the degree of match.

[0101] In step S36, the designer determines whether or not to implement the technology in a product based on whether the false detection rate, accuracy rate, and processing speed are above target values. Setting the threshold value adjusted in step S35 too strictly reduces the false detection rate but also reduces the accuracy rate, whereas setting the threshold value too leniently increases the accuracy rate but also increases the false detection rate, creating a trade-off. The threshold value is determined after taking this trade-off into consideration.

[0102] If the false positive rate is set to a target value of less than 1%, for example, and if thick paper is mistakenly detected as plain paper or thin paper in more than one case out of 100, the process returns to step S31 and starts again from collecting training data 2010. In addition to the false positive rate, the accuracy rate is also determined, and target values ​​are set for each paper thickness, such as a probability of detecting thick paper as thick paper of 80% or more, or a probability of detecting thin paper as thin paper of 70% or more. If these target values ​​are not met, the process returns to step S31. Furthermore, with regard to processing speed, a target value is set, for example, for the time from acquisition of image data to generation of a match by inference of 100 ms or less on average, and 150 ms or less at most. If this target value is not met, the process similarly returns to step S31. Finally, if the false positive rate, accuracy rate, and processing speed all meet their target values, the generation of a learning model 2000 that can be implemented in a product is completed.

[0103] FIG. 13 is a diagram showing an example of the threshold value of the degree of match and the estimation result. FIG. 13(A) is an example of setting the threshold value of the degree of match. As shown in FIG. 13(A), it is possible to select a setting level (normal or strict) for the degree of match. The setting level is selected on a document thickness determination level setting screen 1100 in FIG. 14. Details of the document thickness determination level setting screen 1100 in FIG. 14 will be described later. The threshold value is higher for the setting level "strict" than for the setting level "normal".

[0104] FIG. 13(B) shows the results of estimating the degree of agreement of document information (here, only paper thickness) for certain images A to C using learning model 2000. Image A has the highest degree of agreement for thick paper at 0.96, exceeding the first and second thresholds at both the "normal" and "strict" setting levels. Image B has the highest degree of agreement for plain paper at 0.7, exceeding only the first threshold at the "normal" setting level. Image C does not have a high degree of agreement for any paper thickness, and does not exceed the first and second thresholds at either the "normal" or "strict" setting level for any paper thickness.

[0105] 15 is a flowchart showing an example of screen transitions according to this embodiment. Screen display and operation are performed using the display unit 905 and operation unit 902 in FIG.

[0106] In step S51, the user turns on the power of the image forming unit 1. In step S52, the display unit 905 of the image forming unit 1 displays a startup screen. When the startup process is complete, in step S53, the display unit 905 of the image forming unit 1 displays a home screen. On the home screen, an initial setting button is arranged together with mode selection buttons such as "copy" and "scan."

[0107] When the initial setting button is pressed in step S54, the display unit 905 of the image forming unit 1 displays an initial setting screen in step S55. The initial setting screen is a screen on which various settings that should be set in advance for operations such as copying and scanning in the image forming unit 1 can be made.

[0108] The initial setting screen has buttons such as "Document Thickness Setting," "Automatic Document Thickness Detection Setting," and "Document Thickness Judgment Level Setting," and by pressing each button, the screen can be displayed.

[0109] If the document thickness setting button on the initial setting screen is pressed in step S56, the process proceeds to step S57, where the document thickness setting screen 1200 shown in FIG. 16 is displayed on the display unit 905 of the image forming unit 1. The document thickness setting screen 1200 allows the user to select the document thickness setting from standard, thick paper, or thin paper. For each paper thickness, optimal paper feed and transport control amounts are preset to reduce the risk of skew or paper feed problems. The user sets the thickness of the document to be loaded on the ADF 51 on the document thickness setting screen 1200 before starting document scanning. If there is no error or omission in the setting, the learning model 2000 does not reset the document thickness. In step S58, if the decision button on the document thickness setting screen 1200 is pressed, the display on the display unit 905 of the image forming unit 1 returns to the initial setting screen.

[0110] If the document thickness detection automatic setting button on the initial setting screen is pressed in step S56, the process proceeds to step S59, where the document thickness detection automatic setting screen 1300 of FIG. 17 is displayed on the display unit 905 of the image forming unit 1. The document thickness detection automatic setting screen 1300 of FIG. 17 allows the user to select either "Automatic Setting" or "No Automatic Setting." If the user selects "Automatic Setting" on the document thickness detection automatic setting screen 1300 of FIG. 17, the document thickness is automatically set without prompting the user as to whether or not to reset the document thickness, as shown in the document thickness reset notification screen 1000 of FIG. 8. If the user selects "No Automatic Setting" on the document thickness detection automatic setting screen 1300 of FIG. 17, the user is prompted as to whether or not to reset the document thickness, as shown in the document thickness reset notification screen 1000 of FIG. 8. If the confirm button on the document thickness detection automatic setting screen 1300 is pressed in step S60, the display on the display unit 905 of the image forming unit 1 returns to the initial setting screen.

[0111] In step S56, if the document thickness determination level setting button on the initial setting screen is pressed, the process proceeds to step S61, and the document thickness determination level setting screen 1100 of FIG.

[0112] 14 allows the user to change the setting level of the threshold for the degree of match. If the user sets a stricter setting level for the threshold for the degree of match (the document thickness setting will not be changed unless the degree of match is closer to 1), the user can set the document thickness setting to be changed or the document thickness reset notification screen 1000 to be displayed only when the paper thickness can be determined with a high probability. In step S62, when the enter button on the document thickness determination level setting screen 1100 is pressed, the display on the display unit 905 of the image forming unit 1 returns to the initial setting screen.

[0113] The settings made on the document thickness reset notification screen 1000, document thickness determination level setting screen 1100, document thickness setting screen 1200, and document thickness detection automatic setting screen 1300 are stored in the ROM of the main body control unit 901 of the image forming unit 1 when they are confirmed, and are retained even after the power is turned off.

[0114] Since the document thickness setting screen 1200 in FIG. 16 is likely to be used frequently by users, a separate document thickness setting button may be displayed to transition from the copy mode or scan mode screen to the document thickness setting screen 1200.

[0115] FIG. 18 is a flowchart showing an example of a procedure for changing the document thickness setting and a reading operation using a learning model.

[0116] In step S71, the user places a document to be copied or scanned and presses the copy start key 158 or scan start button to start the first step in the ADF 51. In step S72, the ADF 51 starts the document feeding, transporting, and reading operation as the first step.

[0117] In step S73, the ADF 51 feeds a document, and after the trailing edge of the document passes the document set sensor 63, the document set sensor 63 determines whether the next document is loaded on the document tray. If there is no next document, there is no need to reset the document thickness, and therefore the ADF 51 does not estimate the degree of match for each document thickness using the learning model 2000.

[0118] If there is a next document, in step S74, the controller 904 of the ADF 51 inputs image data scanned by an image reading unit installed in the ADF 51 or the image forming unit 1, and begins estimating the degree of match with each paper thickness using the learning model 2000. In step S75, the controller 904 of the ADF 51 determines whether or not the estimation of the degree of match has been completed. If it is not determined in step S75 that the estimation has been completed, the controller 904 of the ADF 51 returns to the processing of step S72 and continues the document feed, transport, and reading operation with the same settings. If it is determined in step S75 that the estimation has been completed, the controller 904 of the ADF 51 proceeds to the processing of step S76.

[0119] In step S76, the controller 904 of the ADF 51 determines whether the degree of match for any of the paper thicknesses for which estimation has been completed exceeds a preset first threshold. If the degree of match for any of the paper thicknesses exceeds the preset first threshold, the process proceeds to step S77. If the degree of match for any of the paper thicknesses does not exceed the preset first threshold, the process returns to step S72, the first step is continued, and the next document is read.

[0120] In step S77, the controller 904 of the ADF 51 compares the paper thickness that exceeded the first threshold with the currently set paper thickness and determines whether the paper thickness setting is different. If the paper thickness setting is the same, there is no need to reset the paper thickness, so the controller 904 of the ADF 51 returns to step S72, continues the first step, and performs the scanning operation of the next document. If the paper thickness setting is different, it determines that there is a high possibility that the user has set the document paper thickness incorrectly, and the controller 904 of the ADF 51 proceeds to step S78.

[0121] In step S78, the controller 904 of the ADF 51 determines whether the degree of agreement for any of the paper thicknesses exceeds a preset second threshold. If the degree of agreement for any of the paper thicknesses exceeds the preset second threshold, the process proceeds to step S82, where the paper thickness is reset regardless of automatic resetting of the paper thickness, and the second step is initiated. The controller 904 of the ADF 51 initiates paper feeding and reading operations with a transport control amount that is optimal for the changed document thickness setting. If the degree of agreement for any of the paper thicknesses does not exceed the preset second threshold, the process proceeds to step S79.

[0122] In step S79, the controller 904 of the ADF 51 determines whether automatic resetting of the paper thickness has been instructed. If automatic resetting of the paper thickness has been instructed, the controller 904 of the ADF 51 proceeds to step S82, resets the paper thickness, and starts the second step. The controller 904 of the ADF 51 starts paper feeding and reading operations with a transport control amount that is optimal for the changed document thickness setting.

[0123] If automatic resetting of the paper thickness is not instructed in step S79, the controller 904 of the ADF 51 proceeds to step S80. In step S80, the controller 904 of the ADF 51 temporarily stops the paper feeding and reading operations, and causes the display unit 905 to display an original thickness reset notification screen 1000 as shown in Fig. 8, prompting the user for confirmation and prompting the user to input whether or not to change the original thickness setting.

[0124] In step S81, the controller 904 of the ADF 51 determines whether the document thickness setting has been changed by the user. If the user has not changed the document thickness setting, the controller 904 of the ADF 51 returns to step S72, continues the first step, and performs the reading operation of the next document. If the user has changed the document thickness setting, the controller 904 of the ADF 51 proceeds to step S82, and starts paper feeding and reading operations with the transport control amount that is optimal for the changed document thickness setting.

[0125] FIG. 18 illustrates a flowchart from the first step to the second step. However, it is possible that after the transition to the second step, documents of different thicknesses may be mixed. Even after changing the document thickness setting in the second step, the controller 904 of the ADF 51 may return to the start of the flowchart shown in FIG. 18 and repeatedly start the first step, which estimates the degree of match for each document thickness using the learning model 200. In this embodiment, the first step represents a paper feed and reading operation that includes a process for estimating the degree of match. The first step involves transporting documents with a transport control amount that is applicable to all document classifications. The second step represents a paper feed and reading operation that does not include a process for estimating the degree of match. The second step involves transporting documents with a transport control amount that is optimal for the changed document thickness setting.

[0126] Furthermore, if it is determined in step S77 that the paper thickness settings are different multiple times (for example, three or more times) within the same job, it may be determined that mixed paper thicknesses are being loaded, and control may be performed such that no further estimation or determination of the degree of paper thickness consistency is performed.

[0127] Fig. 19 is an explanatory diagram of an example of optimal transport parameters for each paper thickness. The paper feed linear speed of the transport parameters in Fig. 19 is the maximum linear speed during transport from the start of separation and feeding until the leading edge of the document reaches pull-out driven roller 87 in the configuration diagram of Fig. 5.

[0128] The pull-out linear speed is the maximum linear speed at which the leading edge of the document is transported from the pull-out driven roller 87 until it reaches the pair of reading entrance rollers 89 and 90. The first abutting amount is the abutting amount of the document against the pull-out driven roller 87. The second abutting amount is the abutting amount of the document against the pair of reading entrance rollers 89 and 90.

[0129] Each paper thickness (heavy paper, regular paper, and thin paper) has its own optimal parameters determined by the designer through simulations and paper feed evaluations. In addition, parameters applicable to all paper thicknesses are set for the "general purpose" paper thickness. The "general purpose" paper thickness has low productivity but is designed to minimize skew and damage to the original. For example, the transport parameters for the first step in Figure 18 use the "general purpose" paper thickness parameters.

[0130] The reason why the parameters in Figure 19 are optimal for each paper thickness is that, for example, thick paper has a large mass and requires a larger abutment amount to correct the skew amount, so the first abutment amount and second abutment amount are increased compared to other paper thicknesses.

[0131] Regarding the paper feed linear speed, the faster the better to maintain productivity. However, with thin paper, the linear speed is fast when abutting, and if the abutment amount is large, the leading edge is likely to break, so the pull-out linear speed is slower than for other paper thicknesses, and the first abutment amount and second abutment amount are reduced.

[0132] In this embodiment, paper thickness (heavy paper, plain paper, thin paper) is described as a number of classifications of documents, but in addition to paper thickness, paper type (plain paper, coated paper, non-carbon paper) can also be considered, and in the case of special documents such as coated paper or non-carbon paper, control such as lowering the linear speed compared to plain paper is considered to be the optimal transport control. Also, the checks and business cards used in the explanation of this embodiment are merely examples.

[0133] In this embodiment, size information of the document set on the document tray of the ADF 51 is acquired from, for example, a user input or sensor information of the document tray, and the type of the document (paper thickness, etc.) is estimated from the size information.

[0134] Furthermore, in this embodiment, the image data read by the ADF 51 is used as input, and the type of document (such as paper thickness) is estimated by the machine-learned learning model 2000. Even if the type of document to be read that is actually placed on the document tray differs from the type of document that was set in advance, it is possible to switch to transport control and reading control appropriate for that type of document.

[0135] Each function of the above-described embodiments can be realized by one or more processing circuits. In this specification, the term "processing circuit" includes a processor programmed to perform each function by software, such as a processor implemented by an electronic circuit, and devices such as an ASIC (Application Specific Integrated Circuit), a DSP (Digital Signal Processor), an FPGA (Field Programmable Gate Array), or a conventional circuit module designed to perform each function described above.

[0136] Although the present invention has been described above based on the embodiments, the present invention is not limited to the requirements shown in the above embodiments. These requirements can be changed without departing from the spirit of the present invention, and can be appropriately determined depending on the application form.

[0137] For example, aspects of the present invention are as follows. <1> A document tray, a document transport means for transporting a document from the document tray; a manuscript information acquisition means for acquiring manuscript information; a control amount determining means for determining a transport control amount when transporting the document based on a predetermined classification of the document; an estimation means for estimating a classification of the document by inputting information about the document; and The control amount determining means changes the transport control amount when the classification estimated by the estimating means differs from the predetermined classification of the document. A manuscript processing device characterized by:

[0138] The aforementioned <1> According to the above, if the preset document thickness setting (such as thick paper mode) differs from the document thickness setting estimated from size information, etc., the user may have made an incorrect setting. Therefore, by changing the paper thickness setting, even if the user has made an incorrect setting, reading can be performed with optimal control for each paper thickness. <2> an image reading means for reading an image of the document while the document is being transported by the document transport means; and the estimation means has a learning model that receives information about the document and image data acquired by the image reading means as input and outputs a degree of agreement with a plurality of classifications related to the document; The control amount determining means changes the transport control amount when the degree of coincidence output by the learning model exceeds a threshold. Characterized by <1> The document processing device described above.

[0139] The aforementioned <2> According to the method, if the preset document thickness setting differs from the document thickness setting estimated by the learning model from the image data of the scanned document, the user may have made an incorrect setting. Therefore, by changing the paper thickness setting, even if the user makes an incorrect setting, reading can be performed with optimal control for each paper thickness. <3> notification means for notifying the user; an input means for receiving instructions from a user; and the notification means notifies a user whether or not to change the transport control amount when the degree of coincidence output by the learning model exceeds a threshold; The input means receives an instruction from a user as to whether or not to change the transport control amount. Characterized by <2> The document processing device described above.

[0140] The aforementioned <3> According to the article, since estimations made by a learning model are not necessarily correct, the paper thickness setting can be changed after the user has confirmed that the estimated results are indeed correct. <4> the document information acquisition means is a detection means for detecting the size of the document, The information about the document includes information about the size of the document. Characterized by <1> ~ <3> 10. The document processing apparatus according to claim 9, wherein the document processing apparatus is a document processing apparatus for processing a document.

[0141] The aforementioned <4> According to the method, by adding the size of the document detected by the sensor to the training data, the degree of match can be evaluated with higher accuracy. <5> the document information acquisition means is means for receiving a designation of the document size from a user; The information about the document includes information about the size of the document. Characterized by <1> ~ <4> 10. The document processing apparatus according to claim 9, wherein the document processing apparatus is a document processing apparatus for processing a document.

[0142] The aforementioned <5> According to the method, by adding the size of the document input by the user to the training data, the degree of match can be evaluated with higher accuracy. <6> The transport control amount changed by the control amount determining means is the transport speed of the document or the amount of contact of the document with the document transport means. Characterized by <1> ~ <5> 10. The document processing apparatus according to claim 9, wherein the document processing apparatus is a document processing apparatus for processing a document. <7> the transport control amount includes a general transport control amount that is a transport control amount that is applicable to all classifications of documents, The document transport means is a first step of performing conveyance at the general-purpose conveyance control amount; a second step of carrying out transport using the transport control amount changed by the control amount determination means; have <2> The document processing device according to claim 1.

[0143] The aforementioned <7> According to the method, the document is transported and read more carefully, such as by slowing down the transport speed until the image of the first sheet is read and the type of document is determined, and once the optimal mode is determined based on the estimation results, control is performed in that mode from then on, thereby enabling optimal transport and reading of all documents while maintaining reading operations that do not reduce productivity as much as possible. <8> the document information acquisition means is a detection means for detecting whether or not a document is loaded on a document tray, In the first step, if the detecting means detects that there is no document to be subjected to the second step, the estimating means does not estimate the classification of the document. Characterized by <7> The document processing device described above.

[0144] The aforementioned <8> According to this method, if there is no next document, there is no point in estimating the degree of match thereafter, so it is possible to reduce unnecessary processing and process jobs quickly. <9> In the first step and the second step, the learning model outputs a degree of agreement with a plurality of classifications related to the manuscript, and it is determined whether the degree of agreement output by the learning model exceeds a threshold. Characterized by <7> The document processing device described above.

[0145] The aforementioned <9> According to the method, even if the paper is initially determined to be thick paper, if thin paper or the like is mixed in, it is possible to switch to thin paper mode later. <10> In the first step or the second step, if the degree of agreement output by the learning model exceeds a threshold and the change in the transport control amount occurs multiple times, the learning model does not output the degree of agreement thereafter, or does not change the transport control amount even if the degree of agreement exceeds a threshold. Characterized by <9> The document processing device described above.

[0146] The aforementioned <10> According to the paper feed control method, if the transport control needs to be changed many times, it is highly likely that mixed paper thicknesses and paper types are being loaded, so estimation using a learning model is not performed. <11> There are a plurality of thresholds, When the degree of match exceeds a first threshold value but does not exceed a second threshold value, the notification means notifies the user; When the degree of coincidence exceeds both the first threshold value and the second threshold value, the notification means does not notify the user, and the control amount determination means automatically changes the transport control amount. Characterized by <3> The document processing device described above.

[0147] The aforementioned <11> According to the method, there are multiple thresholds, and if the degree of match is not very high, the user is notified, and if the degree of match is high, the threshold is automatically applied without notifying the user, thereby eliminating unnecessary user operations. <12> The threshold value can be changed by the user. Characterized by <2> or <3> The document processing device described above.

[0148] The aforementioned <12> According to this, the user can set the determination of the degree of match to strict, normal, or loose. <13> The document transport means is Control using the general-purpose transport control amount is continued until a learning model that receives image data acquired by the image reading means as an input and outputs a degree of agreement with a plurality of classifications related to the document is output in the first step. Characterized by <7> The document processing device described above.

[0149] The aforementioned <13> According to the method, it may take some time to make a determination based on the degree of match, and in some cases it may be quicker to wait, but in other cases it may be quicker to continue reading the next document in the first step without waiting, and then proceed to the second step once the result is available, so reading can be continued without waiting. <14> The classification of the document is paper thickness information or paper type information. Characterized by <1> ~ <13> 10. The document processing apparatus according to claim 9, wherein the document processing apparatus is a document processing apparatus for processing a document. <15> The document processing device is installed in an image forming device. Characterized by <1> ~ <13> 10. The document processing apparatus according to claim 9, wherein the document processing apparatus is a document processing apparatus for processing a document.

[0150] The aforementioned <15> According to this, the image forming unit 1 may have a CPU that performs inference using the learning model 2000, which allows for cost reduction. <16> A document processing system in which a document processing device and an external device are communicably connected, The document processing device includes: A document tray, a document transport means for transporting a document from the document tray; a manuscript information acquisition means for acquiring manuscript information; a control amount determining means for determining a transport control amount when transporting the document based on a predetermined classification of the document; and The external device is an estimation means for estimating a classification of the document using information on the document received from the document processing device as an input; and The control amount determining means changes the transport control amount when the classification estimated by the estimating means differs from the predetermined classification of the document. A manuscript processing system characterized by:

[0151] The aforementioned <16> According to this, a PC on a network may have a CPU that performs inference using the learning model 2000, thereby reducing costs. Also, the CPU that performs inference using the learning model 2000 may be located on an external cloud server 908, allowing the learning model 2000 to be shared with another system to achieve efficient learning effects. [Explanation of symbols]

[0152] 1 Image forming unit 51 ADF 57 First length sensor 58 Second length sensor 73 Document width sensor 150 scanner 201 4th length sensor 202 Third length sensor 901 Main unit control section 904 Controller 908 Cloud Server [Prior art documents] [Patent documents]

[0153] [Patent Document 1] Japanese Patent Publication No. 2020-112651

Claims

1. A document tray, a document transport means for transporting a document from the document tray; a manuscript information acquisition means for acquiring manuscript information; a control amount determining means for determining a transport control amount when transporting the document based on a predetermined classification of the document; an estimation means for estimating a classification of the document by inputting information about the document; and The control amount determining means changes the transport control amount when the classification estimated by the estimating means differs from the predetermined classification of the document. A manuscript processing device characterized by:

2. an image reading means for reading an image of the document while the document is being transported by the document transport means; and the estimation means has a learning model that receives information about the document and image data acquired by the image reading means as input and outputs a degree of agreement with a plurality of classifications related to the document; The control amount determining means changes the transport control amount when the degree of coincidence output by the learning model exceeds a threshold.

2. The document processing device according to claim 1, wherein:

3. notification means for notifying the user; an input means for receiving instructions from a user; and the notification means notifies a user whether or not to change the transport control amount when the degree of coincidence output by the learning model exceeds a threshold; The input means receives an instruction from a user as to whether or not to change the transport control amount.

3. The document processing device according to claim 2, wherein:

4. the document information acquisition means is a detection means for detecting the size of the document, The information about the document includes information about the size of the document.

3. The document processing device according to claim 1, wherein:

5. the document information acquisition means is means for receiving a designation of the document size from a user; The information about the document includes information about the size of the document.

3. The document processing device according to claim 1, wherein:

6. The transport control amount changed by the control amount determining means is the transport speed of the document or the amount of contact of the document with the document transport means.

3. The document processing device according to claim 1, wherein:

7. the transport control amount includes a general transport control amount that is a transport control amount that is applicable to all classifications of documents, The document transport means is a first step of performing conveyance at the general-purpose conveyance control amount; a second step of carrying out transport using the transport control amount changed by the control amount determination means; 3. The document processing device according to claim 2, further comprising:

8. the document information acquisition means is a detection means for detecting whether or not a document is loaded on a document tray, In the first step, if the detecting means detects that there is no document to be subjected to the second step, the estimating means does not estimate the classification of the document.

8. The document processing device according to claim 7, wherein:

9. In the first step and the second step, the learning model outputs a degree of agreement with a plurality of classifications related to the manuscript, and it is determined whether the degree of agreement output by the learning model exceeds a threshold.

8. The document processing device according to claim 7, wherein:

10. In the first step or the second step, if the degree of agreement output by the learning model exceeds a threshold and the change in the transport control amount occurs multiple times, the learning model does not output the degree of agreement thereafter, or does not change the transport control amount even if the degree of agreement exceeds a threshold.

10. The document processing device according to claim 9, wherein:

11. There are a plurality of thresholds, When the degree of match exceeds a first threshold value but does not exceed a second threshold value, the notification means notifies the user; When the degree of coincidence exceeds both the first threshold value and the second threshold value, the notification means does not notify the user, and the control amount determination means automatically changes the transport control amount.

4. The document processing device according to claim 3, wherein:

12. The threshold value can be changed by the user.

4. The document processing device according to claim 2 or 3, wherein:

13. The document transport means is The control using the general-purpose transport control amount is continued until a learning model that receives the image data acquired by the image reading means as an input and outputs a degree of agreement with a plurality of classifications related to the document is output in the first step.

8. The document processing device according to claim 7, wherein:

14. The classification of the document is paper thickness information or paper type information.

3. The document processing device according to claim 1, wherein:

15. The document processing device is installed in an image forming device.

3. The document processing device according to claim 1, wherein:

16. A document processing system in which a document processing device and an external device are communicably connected, The document processing device includes: A document tray, a document transport means for transporting a document from the document tray; a manuscript information acquisition means for acquiring manuscript information; a control amount determining means for determining a transport control amount when transporting the document based on a predetermined classification of the document; and The external device is an estimation means for estimating a classification of the document using information on the document received from the document processing device as an input; and The control amount determining means changes the transport control amount when the classification estimated by the estimating means differs from the predetermined classification of the document. A manuscript processing system characterized by:

Images