Print medium identification method and print medium identification system

Abstract

To provide a technique capable of specifying the kind of a printing medium having a subtle difference.SOLUTION: A printing medium specification method includes steps of: (a) acquiring first physical property information on a printing medium; (b) acquiring second physical property information on the printing medium different from the first physical property information; (c) inputting the first physical property information to a first discriminator, and thereby acquiring first discrimination information for discriminating the kind of the printing medium; (d) inputting the second physical property information to a second discriminator, and thereby acquiring second discrimination information for discriminating the kind of the printing medium; and (e) specifying the kind of the printing medium using the first discrimination information and the second discrimination information.SELECTED DRAWING: Figure 10

Description

[Technical Field]

[0001] The present disclosure relates to a print medium specification method and a print medium specification system. [Background technology]

[0002] Patent Document 1 discloses a technology for identifying the type of recording medium. In this conventional technology, the type of recording medium is identified by inputting the specular reflection light amount value, the diffuse reflection light amount value, a value related to the basis weight of the recording medium, and a value related to the thickness or density of the recording medium into a machine-learned trained model for paper type identification. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent Publication No. 2021-59451 Summary of the Invention [Problem to be solved by the invention]

[0004] However, with conventional technology, when a single machine learning model is generated using multiple pieces of physical property information, even if a certain piece of physical property information has outstanding features, it may be averaged out, making it difficult to identify types of print media that have subtle differences. [Means for solving the problem]

[0005] According to a first aspect of the present disclosure, there is provided a print medium identification method for identifying the type of print medium, the method comprising the steps of: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the print medium, the second physical property information being different from the first physical property information; (c) acquiring first discrimination information for discriminating the type of the print medium by inputting the first physical property information into a first discriminator configured as a trained machine learning model; (d) acquiring second discrimination information for discriminating the type of the print medium by inputting the second physical property information into a second discriminator configured as a trained machine learning model; and (e) identifying the type of the print medium using the first discrimination information and the second discrimination information. Each of the first discrimination information and the second discrimination information is a similarity for each type of printing medium, and each of the first discriminator and the second discriminator includes a vector neural network having multiple vector neuron layers and is configured to classify the multiple types of printing medium into multiple classes, and each of the first discrimination information and the second discrimination information is a class-specific similarity calculated between a feature spectrum obtained from the output of a specific layer of the machine learning model and a group of known feature spectra created in advance in association with the multiple classes.

[0006] According to a second aspect of the present disclosure, there is provided a print medium identification system that performs a medium identification process to identify the type of print medium. The system includes a memory that stores a first classifier and a second classifier, each configured as a trained machine learning model, and a processor that performs the medium identification process using the first classifier and the second classifier. The processor is configured to perform the following processes: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the print medium that is different from the first physical property information; (c) acquiring first discrimination information that identifies the type of the print medium by inputting the first physical property information into the first classifier; (d) acquiring second discrimination information that identifies the type of the print medium by inputting the second physical property information into the second classifier; and (e) identifying the type of the print medium using the first discrimination information and the second discrimination information. Each of the first discrimination information and the second discrimination information is a similarity for each type of printing medium, and each of the first discriminator and the second discriminator includes a vector neural network having multiple vector neuron layers and is configured to classify the multiple types of printing medium into multiple classes, and each of the first discrimination information and the second discrimination information is a class-specific similarity calculated between a feature spectrum obtained from the output of a specific layer of the machine learning model and a group of known feature spectra created in advance in association with the multiple classes. [Brief explanation of the drawings]

[0007] [Figure 1] FIG. 1 is a block diagram of a printing system according to an embodiment. [Figure 2] FIG. 1 is a block diagram of an information processing device. [Figure 3] FIG. 3 is an explanatory diagram showing the configuration of a first classifier. [Figure 4] FIG. 4 is an explanatory diagram showing the configuration of a second classifier. [Figure 5] 10 is a flowchart showing the processing procedure of a preparation step. [Figure 6] FIG. 10 is an explanatory diagram showing a medium identifier list. [Figure 7] FIG. 4 is an explanatory diagram showing a print setting table. [Figure 8] FIG. [Figure 9] FIG. 2 is an explanatory diagram showing the configuration of a group of known characteristic spectra. [Figure 10] 10 is a flowchart showing the processing procedure of a medium identification / printing process. DETAILED DESCRIPTION OF THE INVENTION

[0008] 1 is a block diagram showing a printing system according to one embodiment. The printing system includes a printer 10, an information processing device 20, a spectral reflectance measuring device 30, and a reflectance distribution measuring device 40.

[0009] The spectral reflectance measuring instrument 30 performs spectral measurement on the printing medium PM used in the printer 10 in an unprinted state to obtain the spectral reflectance R1(λ) as first physical property information. The spectral reflectance R1(λ) indicates the reflectance at multiple wavelengths λ for light incident on the surface of the printing medium PM at one specific incident angle and reflected at one specific reflection angle. The reflectance distribution measuring instrument 40 performs measurement on the printing medium PM used in the printer 10 in an unprinted state to obtain the reflectance distribution R2(θ) as second physical property information. The reflectance distribution R2(θ) indicates the reflectance for light incident on the surface of the printing medium PM at one or more reflection angles and reflected at multiple reflection angles θ for each incident angle. In this embodiment, the reflectance at multiple reflection angles for only one incident angle is used.

[0010] The first and second physical property information related to the printing medium PM may be other types of physical property information than the spectral reflectance R1(λ) and the reflectance distribution R2(θ). For example, the first and second physical property information may each include one or more of the following: spectral reflectance, spectral transmittance, reflectance distribution, an image captured by a visible light camera, thickness, moisture content, weight, coefficient of friction, and an ultrasonic inspection image. The weight is preferably measured per unit area. The image captured by the visible light camera represents the surface texture of the printing medium PM, allowing the type of printing medium PM to be identified based on differences in texture. Furthermore, the ultrasonic inspection image represents the internal structure of the printing medium PM, allowing the type of printing medium PM to be identified based on differences in internal structure. By using these various types of physical property information, the type of printing medium PM can be identified based on various types of physical property information related to the printing medium PM. When using physical property information other than the spectral reflectance R1(λ) and the reflectance distribution R2(θ), appropriate measuring instruments are used. It is preferable that information expressed by a single value, such as thickness, moisture content, weight, and friction coefficient, be combined with other information to form the first physical property information and the second physical property information from two or more types of information. However, the first physical property information and the second physical property information are formed as mutually different pieces of information. "The first physical property information and the second physical property information are mutually different" means that at least one type of information included therein is different.

[0011] As will be described later, the information processing device 20 inputs the spectral reflectance R1(λ) and the reflectance distribution R2(θ) into a first classifier and a second classifier to obtain first and second discrimination information for discriminating the type of printing medium, and then identifies the type of printing medium PM using these discrimination information. The information processing device 20 further controls the printer 10 to perform printing under appropriate printing conditions according to the identified type of printing medium PM.

[0012] 2 is a block diagram showing the functions of the information processing device 20. The information processing device 20 has a processor 110, a memory 120, an interface circuit 130, and an input device 140 and a display device 150 connected to the interface circuit 130. The printer 10, a spectral reflectance measuring instrument 30, and a reflectance distribution measuring instrument 40 are also connected to the interface circuit 130. The processor 110 not only has the function of executing the processes described in detail below, but also has the function of displaying on the display device 150 data obtained by the processes and data generated in the process of the processes.

[0013] The processor 110 operates to implement the functions of the print processing unit 112, the print setting creation unit 114, the learning processing unit 116, and the medium identification processing unit 118. The print processing unit 112 executes printing processing using the printer 10. The print setting creation unit 114 creates print settings appropriate for the type of print medium PM. The medium identification processing unit 118 executes medium identification processing to identify the type of print medium PM. The functions of these units 112, 114, 116, and 118 are implemented by the processor 110 executing a computer program stored in the memory 120. However, the functions of these units 112, 114, 116, and 118 may also be implemented by hardware circuits. The term "processor" in the present disclosure includes such hardware circuits. Furthermore, the processor that executes various processes may be a processor included in a remote computer connected to the information processing device 20 via a network.

[0014] The memory 120 stores classifiers 201 and 202, training data sets TD1 and TD2, a medium identifier list IDL, known characteristic spectrum sets KS1 and KS2, and a print setting table PST. The first classifier 201 is used in a process of acquiring first discrimination information for discriminating the type of print medium according to the spectral reflectance R1(λ) as the first physical property information. The second classifier 202 is used in a process of acquiring second discrimination information for discriminating the type of print medium according to the reflectance distribution R2(θ) as the second physical property information. An example configuration and operation of the classifiers 201 and 202 will be described later. The training data sets TD1 and TD2 are collections of labeled data used for training the classifiers 201 and 202. The medium identifier list IDL is a list in which medium identifiers and physical property information are registered for each type of print medium. The known feature spectrum groups KS1 and KS2 are sets of feature spectra obtained when training data is input again to the trained classifiers 201 and 202. The feature spectra will be described later. The print setting table PST is a table in which print settings suitable for the type of print medium are registered.

[0015] 3 is an explanatory diagram showing the configuration of the first classifier 201. The first classifier 201 is a vector neural network including, in order from the input layer 211 to which the spectral reflectance R1(λ) is input, a convolution layer 221, a primary vector neuron layer 231, a first convolution vector neuron layer 241, and a second convolution vector neuron layer 251 as intermediate layers, and a classification vector neuron layer 261 as an output layer. Of these six layers 211 to 261, the input layer 211 is the lowest layer, and the classification vector neuron layer 261 is the highest layer. In the following description, the layers 221 to 261 are also referred to as the "Conv layer 221," the "PrimeVN layer 231," the "ConvVN1 layer 241," the "ConvVN2 layer 251," and the "ClassVN layer 261," respectively.

[0016] In this embodiment, the input data to the input layer 211 is the spectral reflectance R1(λ), ​​which is a one-dimensional array of data. For example, the spectral reflectance R1(λ) is data obtained by extracting 36 representative values ​​at intervals of 10 nm from data in the range of 380 nm to 730 nm.

[0017] The configuration of the layers 221 to 261 can be described as follows. <Description of the configuration of the first discriminator 201> ·Conv layer 221: Conv[32,6,2] ·PrimeVN layer 231: PrimeVN[26,1,1] ·ConvVN1 layer 241:ConvVN1[20,5,2] ·ConvVN2 layer 251:ConvVN2[16,4,1] ·ClassVN layer 261:ClassVN[n1+1,3,1] Vector dimension VD: VD=16 In the description of each of these layers 221 to 261, the character string before the parentheses is the layer name, and the numbers in the parentheses are, in order, the number of channels, kernel size, and stride. For example, the layer name of the Conv layer 221 is "Conv," the number of channels is 32, the kernel size is 1 x 6, and the stride is 2. In FIG. 3, these descriptions are shown below each layer. The hatched rectangles drawn in each layer represent the kernels used when calculating the output vectors of the adjacent higher layer. In this embodiment, since the input data is data of a one-dimensional array, the kernels also have one-dimensional arrays. Note that the parameter values ​​used in the description of each of the layers 221 to 261 are merely examples and can be changed as desired.

[0018] The Conv layer 221 is a layer composed of scalar neurons. The four upper layers 231 to 261 are layers composed of vector neurons. A vector neuron is a neuron that uses vectors as input and output. In the above description, the dimension of the output vector of each vector neuron is constant at 16. In the following, the term "node" is used as a superordinate concept of scalar neurons and vector neurons.

[0019] FIG. 3 shows the first axis x and second axis y that define the planar coordinates of the node array for the Conv layer 221, and the third axis z that represents depth. It also shows that the sizes of the Conv layer 221 in the x, y, and z directions are 1, 16, and 32. The sizes in the x and y directions are called "resolution." In this embodiment, the resolution in the x direction is always 1. The size in the z direction is the number of channels. These three axes x, y, and z are also used in other layers as coordinate axes that indicate the position of each node. However, in FIG. 3, these axes x, y, and z are omitted from illustration in layers other than the Conv layer 221.

[0020] As is well known, the resolution W1 in the y direction after convolution is given by the following equation: W1=Ceil{(W0-Wk+1) / S} (1) Here, W0 is the resolution before convolution, Wk is the kernel size, S is the stride, and Ceil{X} is a function that rounds up the decimal point of X. The resolution of each layer shown in FIG. 3 is an example in which the resolution of the input data in the y direction is 36, and the actual resolution of each layer is changed appropriately depending on the size of the input data.

[0021] The ClassVN layer 261 has m channels. These channels output classification output values ​​Class1(1) to Class1(m) for m classes. If the maximum value among these classification output values ​​Class1(1) to Class1(m) is equal to or greater than a predetermined threshold, it can be determined that the class associated with this maximum value is the class to which the input data belongs. On the other hand, if the maximum value among the classification output values ​​Class1(1) to Class1(m) is less than the threshold, it can be determined that the class of the input data is unknown. Generally, m is an integer equal to or greater than 2 and is the number of known classes that can be classified using the first classifier 201. It should be noted that instead of the classification output values ​​Class1(1) to Class1(m), it is also possible to determine the class of the input data using a class-specific similarity S1(i), which will be described later.

[0022] FIG. 3 also illustrates subregions Rn in layers 221, 231, 241, 251, and 261. The subscript "n" in subregion Rn is the code for each layer. For example, subregion R221 indicates a subregion in Conv layer 221. A "subregion Rn" is a region in each layer that is identified by a planar position (x, y) defined by the position of the first axis x and the position of the second axis y, and that includes multiple channels along the third axis z. The subregion Rn has dimensions of "Width" × "Height" × "Depth," corresponding to the first axis x, the second axis y, and the third axis z. In this embodiment, the number of nodes included in one "subregion Rn" is "1 × 1 × number of depths," i.e., "1 × 1 × number of channels."

[0023] The first classifier 201 further includes a similarity calculation unit 271 that calculates similarity S1(i). The similarity calculation unit 271 calculates a feature spectrum (described later) from the output of the ConvVN2 layer 251, and calculates similarity S1(i) for each class using the feature spectrum. Here, i is a parameter indicating the class and takes a value from 1 to m.

[0024] In this disclosure, the vector neuron layer used to calculate the similarity S1(i) is also referred to as the "specific layer." As the specific layer, a vector neuron layer other than the ConvVN2 layer 251 may be used, and any number of vector neuron layers greater than or equal to one may be used. The configuration of the feature spectrum and the method of calculating the similarity using the feature spectrum will be described later.

[0025] The vector neural network used in this embodiment is configured based on the same principle as the vector neural network described in Patent Publication No. 2021-189730 disclosed by the applicant of the present disclosure.

[0026] 4 is an explanatory diagram showing the configuration of the second classifier 202. Like the first classifier 201, the second classifier 202 has an input layer 212, intermediate layers including a Conv layer 222, a PrimeVN layer 232, a ConvVN1 layer 242, and a ConvVN2 layer 252, an output layer including a ClassVN layer 262, and a similarity calculation unit 272.

[0027] As can be seen by comparing FIGS. 3 and 4, layers 212 to 262 of the second classifier 202 have the same configuration as layers 211 to 251 of the first classifier 201. Furthermore, ClassVN layer 262, which is the output layer of the second classifier 202, outputs classification output values ​​Class2(1) to Class2(m) for the same m known classes as the first classifier 201. In other words, the second classifier 202 is configured to perform classification processing on the same m types of print media as the first classifier 201. The only difference between the second classifier 202 and the first classifier 201 is that a reflectance distribution R2(θ) is input to input layer 212. In this embodiment, the reflectance distribution R2(θ) is 36 pieces of one-dimensional data including reflectances at 36 reflection angles θ for one specific input angle. However, the number of reflection angles θ can be changed arbitrarily. The similarity calculation unit 272 calculates a feature spectrum from the output of the ConvVN2 layer 252, and calculates a class-specific similarity S2(i) using the feature spectrum.

[0028] 5 is a flowchart showing the processing procedure of the preparation step for the classifiers 201 and 202. This preparation step is a step that is executed by the manufacturer of the printer 10, for example.

[0029] In step S10, first physical property information and second physical property information are acquired for each of the multiple print media. As described above, in this embodiment, the first physical property information is the spectral reflectance R1(λ), ​​which is measured using the spectral reflectance measuring instrument 30. The second physical property information is the reflectance distribution R2(θ), which is measured using the reflectance distribution measuring instrument 40. These measurements are preferably performed at multiple locations on the same print medium. It is also preferable to perform data extension to account for variations in the measurement results. Generally, measurement results vary depending on the date of color measurement and the measuring instrument. Data extension is a process that generates multiple measurement results by adding random variations to the measurement results in order to simulate such variations. Specifically, multiple spectral reflectances R1(λ) are created by adding random variations to the spectral reflectance R1(λ) obtained from a single measurement. The same applies to the reflectance distribution R2(θ). The first physical property information and the second physical property information acquired in step S10 are used as training data sets TD1 and TD2.

[0030] In step S20, a medium identifier list IDL is created for multiple print media. FIG. 6 is an explanatory diagram showing the medium identifier list IDL. The medium identifier list IDL registers the medium identifier assigned to each print medium, the medium name, the class number, the data sub-number, and representative data of the first physical property information and the second physical property information. In this example, m print media are assigned medium identifiers "A-1" to "Am." The medium name is the name of the print medium displayed in the window where the user sets printing conditions. The data sub-number is a number used to distinguish between multiple pieces of physical property information related to the same print medium. In this example, for each print medium, three spectral reflectances R1(λ) are registered as the first physical property information, and three reflectance distributions R2(θ) are registered as the second physical property information. However, the number of pieces of physical property information related to each print medium may differ. It is sufficient for one or more pieces of data to be registered as the first physical property information and the second physical property information for each print medium, but it is preferable that multiple pieces of data be registered. The medium identifier list IDL may also be configured not to include the first physical property information and the second physical property information.

[0031] In step S30 of FIG. 5, print settings are created for multiple print media, and the print setting creation unit 114 registers the print settings in the print setting table PST. FIG. 7 is an explanatory diagram showing the print setting table PST. Each record in the print setting table PST registers a medium identifier and print settings for each print medium. In this example, printer profiles PR1-PRm, medium feed speeds FS1-FSm, and drying times DT1-DTm are registered as print settings. The printer profiles PR1-PRm are output color profiles for the printer 10 and are created for each print medium. Specifically, a printer profile can be created by printing a test chart on a print medium without color correction using the printer 10, subjecting the test chart to spectroscopic measurement using the spectral reflectance measurement instrument 30, and processing the spectroscopic measurement results using the print setting creation unit 114. The medium feed speeds FS1-FSm and drying times DT1-DTm can also be determined experimentally. The "drying time" refers to the time it takes for the print medium to dry after printing in a dryer (not shown) inside the printer 10. In printers that dry the print medium by blowing air onto it after printing, the "drying time" refers to the time it takes for the air to blow. In printers that do not have a dryer, the "drying time" refers to the waiting time for the medium to dry naturally. Note that while other initial items may be set as print settings, it is preferable to create print settings that include at least a printer profile.

[0032] In step S40 of FIG. 5, the user sets parameters for the classifiers 201 and 202. In step S50, the learning processing unit 116 uses the training data sets TD1 and TD2 to train the classifiers 201 and 202, respectively. The first training data set TD1 is a collection of labeled spectral reflectance R1(λ) for m types of printing media. The second training data set TD2 is a collection of labeled reflectance distributions R2(θ) for m types of printing media. The spectral reflectance R1(λ) and reflectance distribution R2(θ) were each measured in an unprinted area. When learning is complete, the trained classifiers 201 and 202 are stored in memory 120.

[0033] In step S60, the learning processing unit 116 again inputs the training data sets TD1 and TD2 into the trained classifiers 201 and 202 to generate known feature spectrum sets KS1 and KS2. The known feature spectrum sets KS1 and KS2 are collections of feature spectra, which will be described below. The following mainly describes a method for generating the known feature spectrum set KS1 associated with the first classifier 201.

[0034] FIG. 8 is an explanatory diagram showing a feature spectrum Sp obtained by inputting arbitrary input data to the trained first classifier 201. Here, the feature spectrum Sp obtained from the output of the ConvVN2 layer 251 will be described. The horizontal axis in FIG. 8 represents the spectral position represented by a combination of the element number ND of the output vector of a node at one planar position (x, y) of the ConvVN2 layer 251 and the channel number NC. In this embodiment, since the vector dimension of the node is 16, the element number ND of the output vector is 16, ranging from 0 to 15. Furthermore, since the number of channels in the ConvVN2 layer 251 is 16, the channel number NC is 16, ranging from 0 to 15.

[0035] The vertical axis of Fig. 8 shows the feature value C at each spectral position. V In this example, the feature value C V is the value of each element of the output vector V ND Note that the feature value C V As the value of each element of the output vector VND Alternatively, the activation value may be used as is. In the latter case, the feature value C included in the feature spectrum Sp is used as the activation value. V The number of nodes is equal to the number of channels, which is 16. The activation value is a value corresponding to the vector length of the output vector of the node.

[0036] The number of feature spectra Sp obtained from the output of the ConvVN2 layer 251 for one piece of input data is equal to the number of planar positions (x, y) of the ConvVN2 layer 251, and is therefore 1×3=3.

[0037] The similarity calculation unit 271 inputs the training data again to the trained first classifier 201 to calculate the feature spectrum Sp shown in FIG. 8, and registers it in the known feature spectrum group KS1.

[0038] FIG. 9 is an explanatory diagram showing the configuration of the known feature spectrum group KS1. Each record in the known feature spectrum group KS1 includes a record number, a layer name, a label Lb, and a known feature spectrum KSp. The known feature spectrum KSp is the same as the feature spectrum Sp in FIG. 8 obtained in response to input of training data. In the example of FIG. 9, known feature spectra KSp associated with each label Lb value are generated and registered from the output of the ConvVN2 layer 251 obtained when the first training data group TD1 is input to the trained first classifier 201. For example, 1_max known feature spectra KSp are registered in association with label Lb=1, 2_max known feature spectra KSp are registered in association with label Lb=2, and m_max known feature spectra KSp are registered in association with label Lb=m. 1_max, 2_max, and m_max are each an integer greater than or equal to 2. As described above, each label Lb corresponds to a different known class. Therefore, it can be understood that each known feature spectrum KSp in the known feature spectrum group KS1 is registered in association with one of the m known classes.

[0039] The known feature spectrum group KS2 associated with the second classifier 202 is also created in the same manner as the known feature spectrum group KS1. Note that the training data group used in step S60 does not need to be the same as the plurality of training data groups TD1 and TD2 used in step S50. However, if some or all of the plurality of training data groups TD1 and TD2 used in step S50 are used in step S60 as well, there is an advantage in that there is no need to prepare new training data.

[0040] 10 is a flowchart showing the processing procedure of the medium identification / printing step using the trained classifiers 201 and 202. This medium identification / printing step is executed by the user of the printer 10, for example.

[0041] In step S110, the medium identification processing unit 118 acquires first physical property information for the target print medium, which is the print medium to be processed, and acquires second physical property information in step S120. As described above, in this embodiment, the first physical property information is the spectral reflectance R1(λ) of the unprinted area, and the second physical property information is the reflectance distribution R2(θ) of the unprinted area.

[0042] In step S130, the medium identification processing unit 118 inputs the first physical property information of the target print medium to the trained first classifier 201 to obtain first discrimination information. As the first discrimination information, either the class-specific similarity S1(i) calculated by the similarity calculation unit 271 or the classification output value Class1(i) output from the output layer, ClassVN layer 261, can be used. This point will be described later.

[0043] In step S140, the medium identification processing unit 118 acquires second discrimination information by inputting the second physical property information of the target print medium to the trained second discriminator 202. As the second discrimination information, either the class-specific similarity S2(i) calculated by the similarity calculation unit 272 or the classification output value Class2(i) output from the output layer, ClassVN layer 262, can be used.

[0044] In step S150, the medium identification processing unit 118 identifies the type of the target print medium using the first discrimination information obtained in step S130 and the second discrimination information obtained in step S140.

[0045] The similarity S1(i) for each class as the first discrimination information can be calculated using, for example, the following equation. S1(i)=max[G{Sp(x,y),KSp(i)}] (2) Here, i is an ordinal number for multiple classes, G{a,b} is a function for calculating the similarity between a and b, Sp(x,y) is the feature spectrum at all planar positions (x,y) obtained according to the input data, KSp(i) is all known feature spectra associated with a specific class i, and max[X] is a logical operation that takes the maximum value of X. As the function G{a,b} for calculating the similarity, for example, a formula for calculating cosine similarity or a formula for calculating similarity according to distance can be used.

[0046] The similarity S1(i) is the maximum value of the similarities calculated between each of the feature spectra Sp(x,y) at all planar positions (x,y) of the ConvVN2 layer 251 and each of all known feature spectra KSp(i) corresponding to a specific class i. Such similarity S1(i) is calculated for each of the m classes i corresponding to the m labels Lb. The similarity S1(i) represents the degree to which the first physical property information of the target print medium is similar to the first physical property information of each class. The similarity S2(i) for each class as the second discrimination information is calculated in the same manner as the similarity S1(i).

[0047] As a method for identifying the type of target print medium using the first discrimination information and the second discrimination information, for example, any of the following methods can be adopted. <Identification method M1> If the maximum value of the class-specific integrated judgment value Sa(i) given by the following formula is equal to or greater than a predetermined judgment threshold, the class corresponding to that maximum value is identified as the type of target print medium. Sa(i)=c1×S1(i)+c2×S2(i) (3a) Here, i is an ordinal number indicating the class, and c1 and c2 are different non-zero weights. This integrated judgment value Sa(i) is a value obtained by adding the similarity S1(i) and similarity S2(i) using different weightings. If the maximum value of the integrated judgment value Sa(i) is less than the judgment threshold, the type of the target print medium is determined to be unknown. This identification method M1 can accurately identify the type of the target print medium using the two similarities S1(i) and S2(i).

[0048] The weights c1 and c2 in equation (3a) above may be set to the same value. However, if the weights c1 and c2 are set to different values, different weights are used for the similarity S1(i) obtained from the first characteristic information and the similarity S2(i) obtained from the second characteristic information, which may enable more accurate identification of the type of print medium based on the two types of characteristic information.

[0049] <Identification method M2> If the maximum value of the class-specific integrated judgment value Sb(i) given by the following formula is equal to or greater than a predetermined judgment threshold, the class corresponding to that maximum value is identified as the type of target print medium. Sb(i)=c1×Class1(i)+c2×Class2(i) (3b) This integrated judgment value Sb(i) is obtained by replacing the similarities S1(i) and S2(i) in the above-mentioned identification method M1 with the classification output values ​​Class1(i) and Class2(i). In this identification method M2, the type of the target print medium can be identified by using the classification output values ​​Class1(i) and Class2(i) as the first discrimination information and the second discrimination information, respectively.

[0050] <Identification method M3> If the integrated judgment value Sc given by the following formula is equal to or greater than a predetermined judgment threshold, the class corresponding to the integrated judgment value Sc is identified as the type of target print medium. Sc=max[S1(i), S2(i)] (3c) This integrated judgment value Sc is the maximum value of the class-specific similarities S1(i) and S2(i). If the integrated judgment value Sc is less than the judgment threshold, the type of the target print medium is determined to be unknown. Even when using this identification method M3, the type of the target print medium can be accurately identified using the two similarities S1(i) and S2(i).

[0051] <Identification method M4> If the integrated determination value Sd given by the following formula is equal to or greater than a predetermined determination threshold, the class corresponding to the integrated determination value Sd is identified as the type of target print medium. Sd=max[Class1(i), Class2(i)] (3d) This integrated judgment value Sd is obtained by replacing the similarities S1(i) and S2(i) in the above-mentioned identification method M3 with the classification output values ​​Class1(i) and Class2(i). When this identification method M4 is used, the type of the target print medium can also be identified by using the classification output values ​​Class1(i) and Class2(i) as the first discrimination information and the second discrimination information, respectively.

[0052] According to the above-described identification methods M1 to M4, it is possible to identify the type of target print medium using the first discrimination information and the second discrimination information. As an example, when two types of print media, black media and specular silver media, were identified according to the above-described identification method M1 using spectral reflectance R1(λ) as the first physical property information and reflectance distribution R2(θ) as the second physical property information, the two media could be distinguished and identified with high accuracy. On the other hand, when discrimination was performed using only the first discriminator 201, it was sometimes impossible to distinguish and identify the black media and the specular silver media. The reason for this is presumably that both the black media and the specular silver media have small diffuse reflection components.

[0053] Once the type of target print medium has been identified in this way, in step S160, the medium identification processing unit 118 determines the medium identifier according to the identified type of target print medium. This process is performed, for example, by referencing the medium identifier list IDL shown in FIG. 6. In step S170, the print processing unit 112 selects print settings according to the medium identifier. This process is performed by referencing the print setting table PST shown in FIG. 7. In step S180, the print processing unit 112 executes printing according to the print settings. Note that if it is determined in step S150 that the type of target print medium is unknown, the process of FIG. 10 ends without performing the processes from step S160 onwards.

[0054] As described above, in this embodiment, the type of target printing medium is identified using first discrimination information obtained by inputting first physical property information of the target printing medium into the first discriminator 201 and second discrimination information obtained by inputting second physical property information of the target printing medium into the second discriminator 202, so that the type can be accurately identified even when there are printing media with subtle differences.

[0055] In the above-described embodiment, the classifiers 201 and 202 are configured using the vector neural network disclosed in Japanese Patent Laid-Open No. 2021-189730. However, instead of this, the capsule network disclosed in U.S. Patent No. 5,210,798 or International Publication No. 2019 / 083553 may be used. Furthermore, instead of a vector neural network, the classifiers 201 and 202 may be configured using a convolutional neural network using scalar neurons. Alternatively, the classifiers 201 and 202 may be configured using other types of machine learning models, such as support vector machines and decision trees.

[0056] Other embodiments: The present disclosure is not limited to the above-described embodiments and can be realized in various forms without departing from the spirit thereof. For example, the present disclosure can also be realized in the following aspects. The technical features in the above embodiments corresponding to the technical features in each aspect described below can be appropriately replaced or combined to solve some or all of the problems of the present disclosure or to achieve some or all of the effects of the present disclosure. Furthermore, if a technical feature is not described as essential in this specification, it can be appropriately deleted.

[0057] (1) According to a first aspect of the present disclosure, there is provided a print medium identification method for identifying the type of print medium, the method comprising: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the print medium, the second physical property information being different from the first physical property information; (c) acquiring first discrimination information for discriminating the type of the print medium by inputting the first physical property information into a first discriminator configured as a trained machine learning model; (d) acquiring second discrimination information for discriminating the type of the print medium by inputting the second physical property information into a second discriminator configured as a trained machine learning model; and (e) identifying the type of the print medium using the first discrimination information and the second discrimination information. This method makes it possible to identify the type of printing medium with high accuracy even when there are subtle differences between the types of printing media.

[0058] (2) In the above-mentioned printing medium identification method, step (e) may include a step of determining the type of the printing medium based on an integrated judgment value obtained by adding the first discrimination information and the second physical property information using different weightings. According to this method, the type of print medium can be identified with high accuracy using both pieces of physical property information.

[0059] (3) In the above print medium identification method, each of the first discrimination information and the second discrimination information may be a similarity for each type of the print medium. According to this method, the type of print medium can be identified with high accuracy using the similarity for each type of print medium.

[0060] (4) In the above-described printing medium identification method, each of the first classifier and the second classifier may include a vector neural network having a plurality of vector neuron layers, and may be configured to classify the plurality of types of printing medium into a plurality of classes, and each of the first discrimination information and the second discrimination information may be a class-specific similarity calculated between a feature spectrum obtained from the output of a specific layer of the machine learning model and a group of known feature spectra created in advance in association with the plurality of classes. According to this method, the similarity for each type of print medium can be calculated appropriately.

[0061] (5) In the print medium identification method, the specific layer may have a configuration in which vector neurons arranged on a plane defined by two axes, a first axis and a second axis, are arranged as multiple channels along a third axis that is oriented in a direction different from the first and second axes. The feature spectrum may be one of (i) a first type of feature spectrum in which multiple element values ​​of an output vector of a vector neuron at one planar position in the specific layer are arranged across the multiple channels along the third axis, (ii) a second type of feature spectrum obtained by multiplying each element value of the first type of feature spectrum by an activation value corresponding to the vector length of the output vector, or (iii) a third type of feature spectrum in which the activation values ​​at one planar position in the specific layer are arranged across the multiple channels along the third axis. According to this method, the characteristic spectrum can be easily obtained.

[0062] (6) In the above-described printing medium identification method, each of the first physical property information and the second physical property information may include one or more of spectral reflectance, spectral transmittance, reflectance distribution, an image captured by a visible light camera, thickness, moisture content, weight, coefficient of friction, and an ultrasonic inspection image. According to this method, the type of print medium can be identified using any of various physical property information related to the print medium.

[0063] (7) In the above printing medium identification method, the first physical property information may be the spectral reflectance, and the second physical property information may be the reflectance distribution including reflectances at multiple reflection angles for one or more incident angles. According to this method, even when the type of print medium cannot be identified by only the spectral reflectance or the reflectance distribution, it is possible to identify the type by using both the spectral reflectance and the reflectance distribution.

[0064] (8) According to a second aspect of the present disclosure, there is provided a print medium identification system that performs a medium identification process for identifying the type of print medium. The system includes a memory that stores a first classifier and a second classifier, each configured as a trained machine learning model, and a processor that performs the medium identification process using the first classifier and the second classifier. The processor is configured to perform the following processes: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the print medium that is different from the first physical property information; (c) acquiring first discrimination information for discriminating the type of the print medium by inputting the first physical property information into the first classifier; (d) acquiring second discrimination information for discriminating the type of the print medium by inputting the second physical property information into the second classifier; and (e) identifying the type of the print medium using the first discrimination information and the second discrimination information.

[0065] The present disclosure may be realized in various forms other than those described above, such as a computer program for implementing the functions of the print medium identification system, or a non-transitory storage medium on which the computer program is recorded. [Explanation of symbols]

[0066] 10... printer, 20... information processing device, 30... spectral reflectance measuring device, 40... reflectance distribution measuring device, 110... processor, 112... print processing unit, 114... print setting creation unit, 116... learning processing unit, 118... medium specific processing unit, 120... memory, 130... interface circuit, 140... input device, 150... display device, 201, 202... classifier, 211, 212... input layer, 221, 222... convolution layer, 231, 232... primary vector neuron layer, 241, 242... first convolution vector neuron layer, 251, 252... second convolution vector neuron layer, 261, 262... classification vector neuron layer (output layer), 271, 272... similarity calculation unit

Claims

1. A method for identifying a type of print medium, comprising: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the print medium, the second physical property information being different from the first physical property information; (c) acquiring first discrimination information for discriminating the type of the printing medium by inputting the first physical property information into a first discriminator configured as a trained machine learning model; (d) acquiring second discrimination information for discriminating the type of the printing medium by inputting the second physical property information into a second discriminator configured as a trained machine learning model; (e) identifying the type of the print medium using the first discrimination information and the second discrimination information; and the first discrimination information and the second discrimination information each represent a similarity for each type of the print medium; each of the first classifier and the second classifier includes a vector neural network having a plurality of vector neuron layers, and is configured to classify the plurality of types of print media into a plurality of classes; A print medium identification method, wherein each of the first discrimination information and the second discrimination information is a class-specific similarity calculated between a feature spectrum obtained from the output of a specific layer of the machine learning model and a group of known feature spectra created in advance in association with the multiple classes.

2. The print medium identification method according to claim 1 , A printing medium identification method, wherein step (e) includes a step of determining the type of the printing medium based on an integrated determination value obtained by adding the first discrimination information and the second discrimination information using different weightings.

3. 3. The print medium identification method according to claim 1, further comprising: A print medium identification method, wherein the identification layer is a vector neuron layer other than the output layer of the machine learning model.

4. The print medium identification method according to claim 1 , The specific layer has a configuration in which vector neurons arranged on a plane defined by two axes, a first axis and a second axis, are arranged as a plurality of channels along a third axis in a direction different from the two axes, The characteristic spectrum is (i) a first type of feature spectrum in which a plurality of element values ​​of an output vector of a vector neuron at one plane position in the specific layer are arranged across the plurality of channels along the third axis; (ii) a second type feature spectrum obtained by multiplying each element value of the first type feature spectrum by an activation value corresponding to the vector length of the output vector; (iii) a third type of feature spectrum in which the activation values ​​at one planar position of the specific layer are arranged across the plurality of channels along the third axis; A print medium specifying method, which is any one of the above.

5. The print medium identification method according to any one of claims 1 to 4, A method for identifying a print medium, wherein each of the first physical property information and the second physical property information includes one or more of spectral reflectance, spectral transmittance, reflectance distribution, an image captured by a visible light camera, thickness, moisture content, weight, coefficient of friction, and an ultrasonic inspection image.

6. The print medium identification method according to claim 5, the first physical property information is the spectral reflectance, A print medium identification method, wherein the second physical property information is the reflectance distribution including reflectances at a plurality of reflection angles for one or more incident angles.

7. A print medium identification system that performs a medium identification process to identify a type of print medium, a memory that stores a first classifier and a second classifier that are each configured as a trained machine learning model; a processor that executes the medium identification process using the first classifier and the second classifier; Equipped with The processor: (a) acquiring first physical property information about the print medium; (b) acquiring second physical property information about the printing medium, the second physical property information being different from the first physical property information; (c) inputting the first physical property information into the first discriminator to obtain first discrimination information for discriminating the type of the printing medium; (d) inputting the second physical property information into the second discriminator to obtain second discrimination information for discriminating the type of the printing medium; (e) a process of identifying the type of the print medium using the first discrimination information and the second discrimination information; is configured to run the first discrimination information and the second discrimination information each represent a similarity for each type of the print medium; each of the first classifier and the second classifier includes a vector neural network having a plurality of vector neuron layers, and is configured to classify the plurality of types of print media into a plurality of classes; A print medium identification system, wherein each of the first discrimination information and the second discrimination information is a class-specific similarity calculated between a feature spectrum obtained from the output of a specific layer of the machine learning model and a group of known feature spectra created in advance in association with the multiple classes.

Images