Detection method, learning method, estimation method, printing method, and detection device

Abstract

To provide a technique for separating a periodic component from a non-periodic component, and detecting distortion of a base material.SOLUTION: A method for detecting distortion of a base material includes: a first distortion data acquisition step (step S16) of acquiring first distortion data indicating distortion of a base material when input data for reference is printed; a second distortion data acquisition step (step S17) of acquiring second distortion data indicating distortion of a base material when input data for evaluation is printed; a shift amount calculation step (step S18) of comparing first distortion data and second distortion data while shifting the distortion data in a longitudinal direction, and calculating a shift amount such that distortion is most synchronous; and a data correction step (step S19) of taking a difference between the second distortion data and the first distortion data shifted based on the shift amount, and acquiring correction data.SELECTED DRAWING: Figure 6

Description

Technical Field

[0001] The present invention relates to a technique for detecting distortion of a base material in a printing apparatus that discharges ink onto the surface of a long strip-shaped base material while conveying the base material in the longitudinal direction.

Background Art

[0002] Conventionally, an inkjet printing apparatus that prints an image on a base material by discharging ink from a plurality of heads while conveying a long strip-shaped base material in the longitudinal direction is known. The inkjet printing apparatus discharges inks of different colors from a plurality of heads. Then, a multicolor image is printed on the surface of the base material by overlapping monochromatic images formed by the inks of each color. A conventional printing apparatus is described in, for example, Patent Document 1.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In this type of printing apparatus, when the base material absorbs ink or when the ink on the base material dries, the base material slightly expands and contracts. As a result, distortion occurs in the printed image formed on the surface of the base material. Due to such expansion and contraction of the base material, the positions of the inks discharged from a plurality of heads may shift. In order to suppress a decrease in print quality due to the expansion and contraction of the base material as much as possible, it is necessary to grasp the distortion of the base material due to the expansion and contraction of the base material.

[0005] However, the distortion of the base material includes not only an aperiodic distortion component caused by the expansion and contraction of the base material due to the ink, but also a periodic distortion component caused by other factors such as the structure of the conveyance mechanism. Therefore, in order to more accurately grasp the distortion of the base material, it is necessary to separate and detect the aperiodic distortion component and the periodic distortion component.

[0006] The present invention has been made in view of such circumstances, and an object thereof is to provide a technique capable of separating and detecting the distortion of a base material into a periodic component and an aperiodic component.

Means for Solving the Problems

[0007] In order to solve the above problems, a first invention of the present application is a method for detecting two-dimensional distortion of a long strip-shaped base material in a printing apparatus that discharges ink onto the surface of the base material while conveying the base material in the longitudinal direction, the method comprising: a) a first distortion data acquisition step of acquiring first distortion data indicating the distortion of the base material when reference input data is printed; b) a second distortion data acquisition step of acquiring second distortion data indicating the distortion of the base material when evaluation input data is printed; c) a shift amount calculation step of comparing the first distortion data and the second distortion data while shifting them in the longitudinal direction and calculating a shift amount at which the distortions are most synchronized; and d) a data correction step of obtaining correction data by taking the difference between the second distortion data and the first distortion data shifted based on the shift amount.

[0008] A second invention of the present application is the detection method of the first invention, wherein each of the step a) and the step b) includes: p1) a step of printing an image on the base material based on input data; p2) a step of imaging the base material printed in the step p1); and p3) a step of calculating distortion based on the captured image captured in the step p2), wherein the input data in the step a) is the reference input data, and the input data in the step b) is the evaluation input data.

[0009] The third invention of the present application is the detection method of the second invention, wherein the reference submission data is image data including only a plurality of marks arranged at a predetermined interval, the evaluation submission data is image data including the plurality of marks arranged at the same interval as the reference submission data and an evaluation pattern or character, and in the step p3), distortion is calculated based on the positions of the marks on the base material in the captured image.

[0010] The fourth invention of the present application is the detection method of the third invention, wherein the mark includes a base figure formed in a first color and a mark figure formed in a second color different from the first color at a position overlapping the base figure.

[0011] The fifth invention of the present application is the detection method of the fourth invention, wherein the first color is white that does not require ink ejection, and the second color is black.

[0012] The sixth invention of the present application is the detection method according to any one of the third invention to the fifth invention, wherein in the step p3), distortion is calculated based on a change in the interval between adjacent marks.

[0013] The seventh invention of the present application is the detection method according to any one of the third invention to the fifth invention, wherein in the step p3), distortion is calculated based on the displacement amount of the mark.

[0014] The eighth invention of the present application is Third Invention the detection method according to any one of the seventh invention, wherein the evaluation pattern or character includes a plurality of regions having different density values or printing rates.

[0015] The ninth invention of the present application is a learning method for generating an estimation model capable of estimating the distortion of the base material by machine learning, including: x1) each step of the detection method according to any one of the first invention to the eighth invention; and x2) a learning step of generating an estimation model capable of estimating the distortion of the base material by machine learning, using the evaluation submission data as an input variable and the correction data as teacher data.

[0016] The tenth invention of the present application is an estimation method for estimating the distortion of a long strip-shaped base material in a printing apparatus that discharges ink onto the surface of the base material while transporting the base material in the longitudinal direction, the method comprising: y1) each step of the learning method of the ninth invention; y2) a data acquisition step of acquiring printing input data, which is image data to be printed; and y3) an estimation step of outputting an estimation result indicating the distortion of the base material by the ink based on the printing input data before printing the printing input data. In the step y3), the printing input data is input to the estimation model generated in the step y1), and the distortion output from the estimation model is used as the estimation result.

[0017] The eleventh invention of the present application is a printing method for printing an image on a long strip-shaped base material in a printing apparatus that discharges ink onto the surface of the base material while transporting the base material in the longitudinal direction, the method comprising: z1) each step of the estimation method of the tenth invention; and z2) a printing step of discharging ink onto the surface of the base material while correcting the discharge position of the ink with respect to the base material based on the estimation result.

[0018] The twelfth invention of the present application is a two-dimensional distortion detection device for a long strip-shaped base material in a printing apparatus that discharges ink onto the surface of the base material while transporting the base material in the longitudinal direction, the device comprising: a first distortion data acquisition unit that acquires first distortion data indicating the distortion of the base material with respect to reference input data; a second distortion data acquisition unit that acquires second distortion data indicating the distortion of the base material with respect to evaluation input data; a shift amount calculation unit that compares the first distortion data and the second distortion data while shifting them in the longitudinal direction and calculates a shift amount at which the distortions are most synchronized; and a data correction unit that obtains correction data by taking the difference between the second distortion data and the first distortion data shifted based on the shift amount.

Advantages of the Invention

[0019] According to the first to twelfth inventions of the present application, the distortion of the base material can be detected by separating it into periodic components and non-periodic components.

Brief Description of the Drawings

[0020]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Embodiments for Carrying Out the Invention

[0021] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0022] <1. Configuration of the Printing Apparatus> FIG. 1 is a diagram showing the configuration of a printing apparatus 1 according to an embodiment of the present invention. This printing apparatus 1 is an apparatus that prints an image on the surface of a base material 9 by discharging ink droplets from a plurality of heads 21 to 24 while conveying the long strip-shaped base material 9. The base material 9 may be printing paper, or may be a resin film. Further, the base material 9 may be a metal foil or a glass base material. As shown in FIG. 1, the printing apparatus 1 includes a conveyance mechanism 10, a printing unit 20, a drying unit 30, a camera 40, and a computer 50.

[0023] The conveyance mechanism 10 is a mechanism that conveys the base material 9 in the conveyance direction along its longitudinal direction. The conveyance mechanism 10 of the present embodiment includes an unwinding unit 11, a plurality of conveyance rollers 12, and a winding unit 13. The base material 9 is fed out from the unwinding unit 11 and conveyed along a conveyance path constituted by a plurality of conveyance rollers 12. Each conveyance roller 12 rotates about an axis extending in a direction perpendicular to the conveyance direction, thereby guiding the base material 9 to the downstream side of the conveyance path. The conveyed base material 9 is collected by the winding unit 13. Further, a tension in the conveyance direction is applied to the base material 9. Thereby, slack and wrinkles of the base material 9 during conveyance are suppressed.

[0024] The printing unit 20 is a processing unit that discharges ink droplets (hereinafter referred to as "ink drops") onto the base material 9 conveyed by the conveyance mechanism 10. The printing unit 20 of the present embodiment includes a first head 21, a second head 22, a third head 23, and a fourth head 24. The first head 21, the second head 22, the third head 23, and the fourth head 24 are arranged at intervals along the conveyance direction of the base material 9. The base material 9 is conveyed below the four heads 21 to 24 with the printing surface facing upward.

[0025] FIG. 2 is a partial top view of the printing apparatus 1 in the vicinity of the printing unit 20. As shown by the dashed lines in FIG. 2, a plurality of nozzles 201 arranged parallel to the width direction of the base material 9 are provided on the lower surfaces of the heads 21 to 24. Each of the heads 21 to 24 discharges ink droplets of each color of K (black), C (cyan), M (magenta), and Y, which are color components of a multicolor image, from the plurality of nozzles 201 toward the upper surface of the base material 9, respectively.

[0026] That is, the first head 21 discharges K-color ink droplets onto the upper surface of the base material 9 at a predetermined position on the conveyance path. The second head 22 discharges C-color ink droplets onto the upper surface of the base material 9 on the downstream side of the discharge position of the first head 21. The third head 23 discharges M-color ink droplets onto the upper surface of the base material 9 on the downstream side of the discharge position of the second head 22. The fourth head 24 discharges Y-color ink droplets onto the upper surface of the base material 9 on the downstream side of the discharge position of the third head 23.

[0027] The drying unit 30 is disposed on the downstream side in the conveyance direction of the heads 21 to 24. The drying unit 30 blows heated gas toward the base material 9 to dry the ink adhering to the base material 9. Thereby, the ink is fixed on the base material 9. Note that the drying unit 30 of the present embodiment blows heated gas from the upper surface side of the base material 9. However, instead of the drying unit 30, other drying mechanisms such as a heating roller that contacts the lower surface side of the base material 9 may be used.

[0028] The camera 40 is an imaging device that captures the printed surface of the base material 9 that has passed through the printing unit 20. The camera 40 is disposed on the downstream side in the conveyance direction of the drying unit 30. The camera 40 is disposed to face the printed surface of the base material 9 on the downstream side of the conveyance path from the four heads 21 to 24. For the camera 40, for example, imaging elements such as CCDs and CMOSs and line sensors in which a plurality are arranged in the width direction are used. The camera 40 acquires a captured image by capturing the printed surface of the base material 9. Then, the camera 40 transmits the obtained captured image to the computer 50.

[0029] The computer 50 is an information processing device for controlling the printing apparatus 1. FIG. 3 is a block diagram showing the connection between the computer 50 and each part of the printing apparatus 1. As conceptually shown in FIG. 3, the computer 50 includes a processor 501 such as a CPU, a memory 502 such as a RAM, and a storage unit 503 such as a hard disk drive. In the storage unit 503, a computer program P for executing the learning process and the printing process described later is stored.

[0030] As shown in FIG. 3, the computer 50 is communicably connected to the above-described transport mechanisms 10, the four heads 21 to 24, the drying unit 30, and the camera 40, respectively. Further, the computer 50 is communicably connected to a server 2 which is an external storage device. In the server 2, reference input data D1, evaluation input data D2, and printing input data D3 are stored. These input data D1, D2, and D3 are image data indicating what kind of printing is to be performed on the base material 9 in the printing apparatus 1. The reference input data D1 and the evaluation input data D2 are used for learning the estimation model M described later. The printing input data D3 is image data to be printed in the printing apparatus 1 in order to obtain a printed matter as a product.

[0031] The computer 50 acquires the input data D1 to D3 from the server 2, and based on the input data D1, controls the operation of the transport mechanism 10 and the four heads 21 to 24. Thereby, the printing process in the printing apparatus 1 proceeds.

[0032] <2. Regarding the functions of the computer> In this printing apparatus 1, four heads 21 to 24 print monochromatic images on the upper surface of a substrate 9 by discharging ink droplets. Then, a multicolor image is formed on the upper surface of the substrate 9 by overlapping the four monochromatic images. Therefore, if the positions of the ink droplets discharged from the four heads 21 to 24 on the substrate 9 are shifted from each other, the image quality of the printed matter will deteriorate. Suppressing such mutual misregistration (misalignment) of the monochromatic images on the substrate 9 within an allowable range is an important factor for obtaining high-quality printed matter.

[0033] When the discharged ink is absorbed by the substrate 9 or when the ink on the substrate 9 dries, the substrate 9 expands and contracts unevenly. When such expansion and contraction of the substrate 9 occur, the printed image formed on the printing surface is distorted. In addition, when the substrate 9 expands and contracts, the above-mentioned misregistration is also likely to occur. For example, if the substrate 9 expands and contracts due to the ink discharged by the first head 21 and the substrate 9 is distorted, the landing position of the ink droplets in the first head 21 and the landing positions of the ink droplets in the second and subsequent heads 22 to 24 are shifted. As a result, misregistration occurs.

[0034] Here, the discharge density of each color of ink differs for each image to be printed, and there may be a region with a low discharge density and a region with a high discharge density within one image. For this reason, the distortion that occurs in the substrate 9 differs depending on what kind of image is printed. Therefore, in order to predict misregistration before printing, it is preferable to create an estimation model that has learned the relationship between the input data of the image to be printed and the distortion of the substrate 9 in order to estimate the distortion of the substrate 9 before printing based on the input data of the image to be printed. For that purpose, it is necessary to prepare learning data for learning the relationship between the input data and the distortion of the substrate 9.

[0035] On the other hand, in a printing apparatus having a transport mechanism 10 such as the printing apparatus 1, in addition to the non-periodic distortion of the substrate 9 caused by the absorption and drying of ink, periodic distortion occurs due to manufacturing errors, assembly errors, and vibrations during driving of each part in the transport mechanism 10. Therefore, when attempting to obtain learning data for creating an estimation model, since not only non-periodic distortion but also periodic distortion is superimposed on the learning data, it becomes difficult to accurately learn the correspondence between the input data and the non-periodic distortion of the substrate 9.

[0036] Therefore, the computer 50 of this printing apparatus 1 obtains learning data D8 used for learning processing by detecting and comparing the distortion of the substrate 9 for each of the reference input data D1 and the evaluation input data D2. Further, the computer 50 creates an estimation model for estimating the expansion and contraction state of the substrate 9 before printing from the input data using the learning data D8. Then, the computer 50 has a function of estimating the expansion and contraction state of the substrate 9 before printing based on the input data and performing printing while correcting the ejection position of the ink droplets using the estimation result. FIG. 4 is a block diagram conceptually showing the function of the computer 50.

[0037] As shown in FIG. 4, the computer 50 includes an input data acquisition unit 51, a distortion measurement unit 52, a data correction unit 53, a learning unit 54, an estimation unit 55, a correction value calculation unit 56, and an operation control unit 57. Each function of the input data acquisition unit 51, the distortion measurement unit 52, the data correction unit 53, the learning unit 54, the estimation unit 55, the correction value calculation unit 56, and the operation control unit 57 is realized by the processor 501 of the computer 50 operating according to the computer program P.

[0038] The input data acquisition unit 51 is a processing unit for acquiring reference input data D1, evaluation input data D2, and printing input data D3. The input data acquisition unit 51 reads the reference input data D1, the evaluation input data D2, and the printing input data D3 from the server 2. Then, the input data acquisition unit 51 delivers the read reference input data D1 to the distortion measurement unit 52 and the operation control unit 57. The input data acquisition unit 51 delivers the read evaluation input data D2 to the learning unit 54 and the operation control unit 57. Also, the input data acquisition unit 51 delivers the read printing input data D3 to the estimation unit 55 and the operation control unit 57.

[0039] The distortion measurement unit 52 is a processing unit for measuring the two-dimensional distortion of the base material 9 based on the first captured image D4 and the second captured image D5 transmitted from the camera 40. The distortion measurement unit 52 inputs the measured first distortion data D6 and second distortion data D7 of the base material 9 to the data correction unit 53.

[0040] The data correction unit 53 is a processing unit for correcting the second distortion data D7 using the first distortion data D6 to obtain learning data D8. The data correction unit 53 inputs the obtained learning data D8 to the learning unit 54.

[0041] The distortion measurement unit 52 and the data correction unit 53 constitute a distortion detection unit 60 that corrects the distortion data (second distortion data D7) for the evaluation input data D2 and detects the non-periodic two-dimensional distortion (learning data D8) of the base material 9 for the evaluation input data D2. The function of the distortion detection unit 60 will be described later.

[0042] The learning unit 54 is a processing unit for learning the relationship between the image data and the non-periodic distortion of the base material 9 when the image data is printed. This printing apparatus 1 performs the learning process by the learning unit 54 before printing the printing input data D3 for obtaining the printed matter as a product.

[0043] During the learning process, the learning unit 54 receives the evaluation input data D2 and the learning data D8. As described above, the learning data D8 is the data obtained by detecting the non-periodic distortion of the base material 9 with respect to the evaluation input data D2. The learning unit 54 uses the evaluation input data D2 as the input variable and the learning data D8 detected by the distortion detection unit 60 as the teacher data, and performs the learning process by a supervised machine learning algorithm. The learning unit 54 repeats such a learning process until a predetermined end condition is satisfied. As a result, the learning unit 54 generates an estimation model M that can output an estimation result indicating the non-periodic distortion of the base material 9 based on the input image data. Details of the learning process will be described later.

[0044] The learning unit 54 uses, for example, deep learning (multi-layer neural network) as the machine learning algorithm. However, the machine learning algorithm used by the learning unit 54 is not limited to deep learning. Instead of deep learning, other machine learning algorithms such as Markov random field (MRF) and Boltzmann machine may be used.

[0045] The estimation unit 55 is a processing unit for estimating the non-periodic distortion of the base material 9 when the print input data D3 is printed based on the print input data D3. The estimation unit 55 performs the estimation process using the estimation model M generated by the learning unit 54. The estimation unit 55 inputs the print input data D3 acquired by the input data acquisition unit 51 into the estimation model M. Then, the distortion of the base material 9 corresponding to the print input data D3 is output from the estimation model M. The estimation unit 55 uses the distortion output from this estimation model M as the estimation result D9.

[0046] The correction value calculation unit 56 is a processing unit for calculating the correction value D10 based on the estimation result D9. The correction value calculation unit 56 calculates the correction value D10 in the direction of canceling the two-dimensional distortion of the base material 9 indicated by the estimation result D9. The calculated correction value D10 is input to the operation control unit 57.

[0047] The operation control unit 57 is a processing unit for controlling the operations of the conveyance mechanism 10 and the four heads 21 to 24.

[0048] Before the learning process by the learning unit 54, when the distortion detection unit 60 creates the learning data D8, the reference input data D1 and the evaluation input data D2 are input to the operation control unit 57. Then, the operation control unit 57 outputs command values based on these input data D1 and D2 to the conveyance mechanism 10 and the four heads 21 to 24. Thereby, the conveyance mechanism 10 and the four heads 21 to 24 are operated to print the images of the reference input data D1 and the evaluation input data D2 on the printing surface of the base material 9. At this time, the camera 40 captures the printed base material 9 to obtain the first captured image D4 and the second captured image D5.

[0049] On the other hand, when printing the printing input data D3, the operation control unit 57 corrects the command value based on the printing input data D3 with the correction value D10. Then, the operation control unit 57 outputs the corrected command value to the conveyance mechanism 10 and the four heads 21 to 24. Thereby, the conveyance mechanism 10 and the four heads 21 to 24 are operated to print the image of the printing input data D3 on the printing surface of the base material 9.

[0050] <Regarding the distortion detection process> Next, the configuration of the distortion detection unit 60 and the distortion detection process executed in the distortion detection unit 60 will be described. The distortion detection device of this embodiment is configured by a computer 50 having a distortion detection unit 60. FIG. 5 is a block diagram conceptually showing the functions of the distortion detection unit 60.

[0051] As shown in FIGS. 4 and 5, the distortion detection unit 60 includes the above-described distortion measurement unit 52 and data correction unit 53. Also, as shown in FIG. 5, the distortion measurement unit 52 includes a first distortion data acquisition unit 61 and a second distortion data acquisition unit 62. The data correction unit 53 includes a shift amount calculation unit 63 and a difference processing unit 64.

[0052] The first distortion data acquisition unit 61 receives the reference input data D1 from the input data acquisition unit 51 and the first captured image D4 from the camera 40. The first captured image D4 is an image captured by the camera 40 of the reference input data D1 printed on the substrate 9 by the printing apparatus 1. The first distortion data acquisition unit 61 calculates the two-dimensional distortion of the substrate 9 shown in the first captured image D4 based on the reference input data D1 and the first captured image D4, and acquires the first distortion data D6. The method for calculating the distortion will be described later.

[0053] The second distortion data acquisition unit 62 receives the reference input data D1 from the input data acquisition unit 51 and the second captured image D5 from the camera 40. The second captured image D5 is an image captured by the camera 40 of the evaluation input data D2 printed on the substrate 9 by the printing apparatus 1. The second distortion data acquisition unit 62 calculates the two-dimensional distortion of the substrate 9 shown in the second captured image D5 based on the reference input data D1 and the second captured image D5, and acquires the second distortion data D7. The method for calculating the distortion will be described later.

[0054] The shift amount calculation unit 63 receives the first distortion data D6 from the first distortion data acquisition unit 61 and the second distortion data D7 from the second distortion data acquisition unit 62. The shift amount calculation unit 63 compares the first distortion data D6 and the second distortion data D7 while shifting them in the conveyance direction (the longitudinal direction of the substrate 9), and calculates the shift amount DS at which the distortions of the first distortion data D6 and the second distortion data D7 are most synchronized. The method for calculating the shift amount DS will be described later.

[0055] The difference processing unit 64 obtains the learning data D8, which is correction data, by taking the difference between the second distortion data D7 and the first distortion data D6 shifted based on the shift amount DS input from the shift amount calculation unit 63.

[0056] FIG. 6 is a flowchart showing the flow of the distortion detection process in the distortion detection unit 60. This distortion detection process is executed before the learning process performed in the learning unit 54.

[0057] As shown in FIG. 6, when performing the distortion detection process, first, the input data acquisition unit 51 reads the reference input data D1 and the evaluation input data D2 from the server 2. As a result, the input data acquisition unit 51 acquires the reference input data D1 and the evaluation input data D2 (step S11). Then, the input data acquisition unit 51 inputs the reference input data D1 to the operation control unit 57 and the distortion measurement unit 52.

[0058] FIG. 7 is a diagram exaggerating an example of the reference input data D1. FIG. 8 is a diagram showing an example of the evaluation input data D2 corresponding to the reference input data D1 of the example in FIG. 7. As shown in FIGS. 7 and 8, a plurality of grid marks 90 are incorporated in the reference input data D1 and each evaluation input data D2, respectively.

[0059] The reference input data D1 is image data including only a plurality of grid marks 90 arranged at a predetermined interval. In the reference input data D1, the plurality of grid marks 90 are arranged at a predetermined interval in the conveyance direction and the width direction. In the examples of FIGS. 7 and 8, for easy understanding, the grid marks 90 are shown large with respect to the sizes of the input data D1 and D2, and the number of the grid marks 90 is small (the density is low).

[0060] The evaluation input data D2 is image data including a plurality of grid marks 90 arranged at the same interval as the reference input data D1 and a pattern 91. The pattern 91 is an evaluation pattern or character for evaluating distortion. A plurality of evaluation input data D2 having different patterns 91 are used.

[0061] The evaluation submission data D2 is used to learn the distortion of the base material 9 later. The amount of expansion and contraction of the base material 9 varies depending on the amount of ink ejected onto the base material 9. Therefore, for at least a part of the evaluation submission data D2, it is desirable that the pattern 91 is an image including a plurality of regions with different density values or printing rates. Thereby, the amount of expansion and contraction of the base material 9 according to the density value or the printing rate can be learned.

[0062] The grid mark 90 is a mark indicating a predetermined coordinate position in the reference submission data D1 and the evaluation submission data D2. The grid marks 90 are arranged throughout the reference submission data D1 and the evaluation submission data D2. Also, the plurality of grid marks 90 are arranged at intervals along the conveyance direction and the width direction of the base material 9. Further, in the evaluation submission data D2, the plurality of grid marks 90 are arranged on the front side of the pattern 91.

[0063] FIG. 9 is an enlarged view of a part of the evaluation submission data D2. In the upper region of FIG. 9, the grid mark 90 is arranged in the pattern 91a filled with black. Also, in the lower region of FIG. 9, the grid mark 90 is arranged in the pattern 91b filled with white.

[0064] As shown in FIG. 9, the grid mark 90 of the present embodiment is composed of a base figure 901 and a mark figure 902. The base figure 901 is a rectangular figure filled with white. White means a density value of 0% and is a color that does not require ink ejection. The mark figure 902 is a black cross-shaped figure overlapping the front side of the base figure 901.

[0065] When the grid mark 90 is arranged on a black background like the upper region of FIG. 9, the black pattern 91a serving as the background and the mark figure 902 are the same black color. However, since the mark figure 902 is on the white base figure 901, the mark figure 902 can be identified. Also, when the grid mark 90 is arranged on the white pattern 91b serving as the background like the lower region of FIG. 9, the base figure 901 and the background are the same white color, so the boundary cannot be identified, but the black mark figure 902 can be identified.

[0066] As described above, in this embodiment, since the grid mark 90 is composed of the white base figure 901 and the black mark figure 902 overlapping the front side of the base figure 901, the mark figure 902 can be identified regardless of the color of the background pattern 91. Also, since the base figure 901 that occupies a large area of the grid mark 90 is white with a density value of 0%, the amount of ink ejected for printing the grid mark 90 can be suppressed. That is, it is possible to suppress the occurrence of distortion in the base material 9 due to the printing of the grid mark 90.

[0067] Subsequent to step S11, the operation control unit 57 controls the operations of the transport mechanism 10 and the four heads 21 to 24 based on the reference input data D1. Thereby, the reference input data D1 is printed on the printing surface of the base material 9 (step S12). That is, a plurality of grid marks 90 are printed on the printing surface of the base material 9. Then, the printing surface of the base material 9 on which the ink has been ejected is dried by the drying unit 30, and the ink is fixed.

[0068] The camera 40 photographs the printing surface of the base material 9 on which the reference input data D1 is printed and acquires the first photographed image D4 (step S13). Then, the first photographed image D4 obtained by the photographing is transmitted from the camera 40 to the computer 50 and input to the first distortion data acquisition unit 61 of the distortion measurement unit 52.

[0069] Further, the operation control unit 57 controls the operations of the transport mechanism 10 and the four heads 21 to 24 based on the evaluation input manuscript data D2. Thereby, the evaluation input manuscript data D2 is printed on the printing surface of the base material 9 (step S14). That is, a pattern 91 and a plurality of grid marks 90 are printed on the printing surface of the base material 9. Then, the printing surface of the base material 9 on which the ink has been ejected is dried by the drying unit 30, and the ink is fixed.

[0070] The camera 40 photographs the printing surface of the base material 9 on which the evaluation input manuscript data D2 is printed, and acquires a second photographed image D5 (step S15). Then, the second photographed image D5 obtained by photographing is transmitted from the camera 40 to the computer 50 and input to the second distortion data acquisition unit 62 of the distortion measurement unit 52.

[0071] The first distortion data acquisition unit 61 analyzes the first photographed image D4 and acquires first distortion data D6 that measures the distortion of the base material 9 on which the reference input manuscript data D1 is printed (step S16). At this time, while referring to the reference input manuscript data D1, the first distortion data acquisition unit 61 extracts the positions of a plurality of grid marks 90 on the base material 9 from the first photographed image D4. Then, based on the positions of those grid marks 90, the distortion of the base material 9 is measured.

[0072] In this embodiment, steps S12, S13, and S16 constitute a first distortion data acquisition step of acquiring first distortion data D6 indicating the distortion of the base material 9 when the reference input manuscript data D1 is printed.

[0073] The second distortion data acquisition unit 62 analyzes the second photographed image D5 and acquires second distortion data D7 that measures the distortion of the base material 9 on which the evaluation input manuscript data D2 is printed (step S17). At this time, while referring to the reference input manuscript data D1, the second distortion data acquisition unit 62 extracts the positions of a plurality of grid marks 90 on the base material 9 from the second photographed image D5. Then, based on the positions of those grid marks 90, the distortion of the base material 9 is measured.

[0074] In addition, in the present embodiment, a second distortion data acquisition step is configured to acquire second distortion data D7 indicating the distortion of the base material 9 when printing the evaluation input data D2 by steps S14, S15, and S17.

[0075] In step S17, the position of the grid mark 90 in the reference input data D1 is referred to for extracting the position of the grid mark 90, but the position of the grid mark 90 in the evaluation input data D2 may also be referred to. In the present embodiment, since the position of the grid mark 90 in the reference input data D1 is the same as the position of the grid mark 90 in the evaluation input data D2, by referring to the position of the grid mark 90 in the reference input data D1 in step S17, the steps common to step S15 can be omitted.

[0076] However, the position of the grid mark 90 in the reference input data D1 and the position of the grid mark 90 in the evaluation input data D2 may be different. In that case, in step S16, the position of the grid mark 90 in the evaluation input data D2 is referred to for extracting the position of the grid mark 90.

[0077] Here, while referring to FIGS. 10 and 11, a method for the first distortion data acquisition unit 61 and the second distortion data acquisition unit 62 (hereinafter collectively referred to as the distortion measurement unit 52) to measure the distortion of the first captured image D4 or the second captured image D5 in steps S16 and S17 will be described.

[0078] FIG. 10 is a diagram showing an example of a method for measuring the distortion of the base material 9. In the example of FIG. 10, the distortion measurement unit 52 measures the interval between adjacent grid marks 90 in the captured images D4 and D5. Then, the difference between the measured interval of the grid marks 90 and the interval of the grid marks 90 in the reference input data D1 is calculated as the distortion. That is, in the method of FIG. 10, the change in the interval between adjacent grid marks 90 is measured. The distortion measurement unit 52 performs such distortion measurement for all adjacent grid marks 90 in the captured images D4 and D5. As a result, a heat map showing the distribution of distortion in the base material 9 is obtained. That is, in the method of FIG. 10, the heat map in the first captured image D4 becomes the first distortion data D6, and the heat map in the second captured image D5 becomes the second distortion data D7.

[0079] When the base material 9 is distorted due to expansion and contraction, the displacement amount of each region on the base material 9 is not only the expansion and contraction amount of that region but also affected by the expansion and contraction amounts of other regions, and becomes a value obtained by accumulating those expansion and contraction amounts. However, the heat map obtained by the measurement method of FIG. 10 represents the local distortion of each region of the base material 9. Therefore, the heat map is easy to handle as teacher data in the learning process described later.

[0080] FIG. 11 is a diagram showing another example of a method for measuring the distortion of the base material 9. In the example of FIG. 11, the distortion measurement unit 52 measures the positions of the respective grid marks 90 in the photographed images D4 and D5. Specifically, with a specific grid mark 90 in the photographed images D4 and D5 as the origin, the coordinate positions of the other grid marks 90 with respect to the origin are measured respectively. Then, the difference between the measured coordinate positions and the coordinate positions of the grid marks 90 in the reference input data D1 is calculated as the distortion. That is, in the method of FIG. 11, the displacement amount of each grid mark 90 on the base material 9 is measured. The distortion measurement unit 52 performs such distortion measurement for all the grid marks 90 in the photographed images D4 and D5. As a result, a vector map showing the distribution of the distortion on the base material 9 is obtained. That is, in the method of FIG. 11, the vector map in the first photographed image D4 becomes the first distortion data D6, and the vector map in the second photographed image D5 becomes the second distortion data D7.

[0081] The vector map obtained by the measurement method of FIG. 11 represents the displacement amount of the coordinate positions of each part of the base material 9. In the learning process performed after the distortion detection process, when such a vector map is used as learning data, the estimation result output from the estimation model M described later also represents the displacement amount of the coordinate positions of each part of the base material 9. Therefore, if the measurement method of FIG. 11 is adopted, it becomes easier to use the estimation result when the correction value calculation unit 56 corrects the coordinate position of the ink ejected onto the base material 9.

[0082] Thereafter, the first distortion data acquisition unit 61 delivers the first distortion data D6 to the shift amount calculation unit 63 of the data correction unit 53. Also, the second distortion data acquisition unit 62 delivers the second distortion data D7 to the shift amount calculation unit 63.

[0083] Subsequently, the shift amount calculation unit 63 compares the first distortion data D6 and the second distortion data D7 while shifting them in the conveyance direction (the longitudinal direction of the base material 9), and calculates the shift amount DS at which the distortions of the first distortion data D6 and the second distortion data D7 are most synchronized (step S18: shift amount calculation step).

[0084] FIG. 12 conceptually shows a state when the shift amount DS is calculated from the first strain data D6 and the second strain data D7. FIG. 12 shows examples of the reference input data D1, the evaluation input data D2, the first strain data D6, and the second strain data D7. In the example of FIG. 12, pattern 91 of the evaluation input data D2 is such that half in the width direction is filled with black and the other half is filled with white.

[0085] As shown in FIG. 12, in both the first strain data D6 and the second strain data D7, strain appears periodically in the conveyance direction. Here, since the ink ejection amount in the reference input data D1 is small, the first strain data D6 hardly includes the aperiodic strain of the base material 9 due to ink ejection. For this reason, the strain of the base material 9 appearing in the first strain data D6 is considered to be a periodic strain not caused by ink ejection.

[0086] On the other hand, since the plurality of prepared evaluation input data D2 includes various patterns 91, an aperiodic strain is generated in the base material 9 due to ink ejection during printing. For this reason, in the second strain data D7, the periodic strain appearing in the first strain data D6 and the aperiodic strain due to ink ejection are superimposed.

[0087] Therefore, in step S18, as shown in FIG. 12, the shift amount calculation unit 63 creates a plurality of shift data D6s obtained by shifting the first strain data D6 in the conveyance direction while changing the shift amount DS, and compares it with the second strain data D7. Then, the shift amount DS at which the strains of the first strain data D6 and the second strain data D7 are most synchronized is calculated. In the present embodiment, in the shift data D6s and the second strain data D7, the shift amount DS that is most approximated using the least squares method is selected. Specifically, the difference for each corresponding position of the heat map is taken, and the shift amount DS at which the sum of the squares of the differences is minimized is selected.

[0088] Thereafter, the difference processing unit 64 takes the difference between the second distortion data D7 and the first distortion data D6 shifted based on the shift amount DS calculated by the shift amount calculation unit 63, and corrects the second distortion data D7 (step S19: data correction step). In this way, the difference processing unit 64 acquires the learning data D8, which is the correction data.

[0089] By correcting the second distortion data D7 in this way, the periodic distortion components included in the second distortion data D7 can be removed, and the learning data D8 including only the aperiodic distortion components can be obtained.

[0090] <4. Regarding the learning process> Subsequently, the learning process executed in the learning unit 54 will be described with reference to FIG. 13. FIG. 13 is a flowchart showing the flow of the learning process.

[0091] In the learning process, first, the evaluation input data D2 is input to the learning unit 54 from the input document data acquisition unit 51, and the learning data D8 corresponding to the evaluation input data D2 is input from the difference processing unit 64 of the data correction unit 53 (step S21).

[0092] The learning unit 54 performs machine learning using the evaluation input data D2 as the input variable and the learning data D8 as the teacher data by a supervised machine learning program. At this time, the learning unit 54 prepares an estimation model M for estimating the aperiodic distortion of the base material 9 based on the image data. The estimation model M outputs an estimation result of the aperiodic distortion based on the input evaluation input data D2. The learning unit 54 adjusts the parameters of the estimation model M so that the estimation result output from the estimation model M approaches the learning data D8, which is the teacher data (step S22).

[0093] After that, the learning unit 54 determines whether or not a predetermined end condition is satisfied (step S23). For example, the end condition may be that the difference between the estimation result and the teacher data becomes smaller than a preset threshold value. Also, the end condition may be that the number of repetitions of steps S21 to S22 reaches a preset threshold value. When the end condition is not satisfied (step S23: No), the computer 50 repeats the above-described processing of steps S21 to S22. At this time, there are a plurality of evaluation input data D2 and learning data D8, and different evaluation input data D2 and learning data D8 may be used each time the processing is repeated.

[0094] By repeating the learning process of steps S21 to S22, the estimation accuracy of the estimation model M is improved. Eventually, when the end condition is satisfied (step S23: Yes), the learning unit 54 ends the learning process. As a result, a learned estimation model M that can accurately estimate the non-periodic distortion of the base material 9 when the input image data is printed is generated. The learning unit 54 provides the generated estimation model M to the estimation unit 55.

[0095] <5. Regarding the printing process> Subsequently, the printing process executed in the printing apparatus 1 after the above-described learning process will be described with reference to FIG. 14. FIG. 14 is a flowchart showing the flow of the printing process.

[0096] As shown in FIG. 10, when performing the printing process, first, printing input data D3 to be printed is acquired (step S31). Specifically, the input data acquisition unit 51 reads the printing input data D3 from the server 2. Then, the input data acquisition unit 51 inputs the printing input data D3 to the estimation unit 55 and the operation control unit 57.

[0097] The estimation unit 55 inputs the printing input data D3 into the estimation model M generated by the learning unit 54. Then, the estimation model M outputs an estimation result D9 of the distortion of the substrate 9 (step S32). This estimation result D9 indicates the estimated value of the aperiodic distortion of the substrate 9 by the ink when the printing input data D3 is printed in the printing apparatus 1. The estimation unit 55 outputs the obtained estimation result D9 to the correction value calculation unit 56.

[0098] Based on the estimation result D9 output from the estimation unit 55, the correction value calculation unit 56 calculates a correction value D10 (step S33). This correction value D10 is a control value for finely adjusting the ejection position of the ink droplets onto the substrate 9. The correction value calculation unit 56 sets the correction value D10 in a direction to cancel the distortion of the substrate 9 indicated by the estimation result D9. For example, when it is estimated that a certain part of the substrate 9 is displaced to one side in the width direction due to the distortion of the substrate 9, the correction value calculation unit 56 calculates the correction value D10 so as to correct the ejection position of the ink droplets to the other side in the width direction. Then, the correction value calculation unit 56 inputs the calculated correction value D10 to the operation control unit 57.

[0099] Thereafter, based on the printing input data D3 acquired from the input data acquisition unit 51 and the correction value D10 acquired from the correction value calculation unit 56, the operation control unit 57 controls the operation of the conveyance mechanism 10 and the four heads 21 to 24. The operation control unit 57 corrects the ejection position of the ink specified by the printing input data D3 according to the correction value D10. This correction is performed, for example, for each pixel of the printing input data D3. Then, ink droplets are ejected onto the corrected ejection position on the printing surface of the substrate 9. Thereby, the printing input data D3 is printed on the printing surface of the substrate 9 (step S34).

[0100] As described above, in this printing apparatus 1, before printing the input data D3 for printing, based on the input data D3 for printing, the non-periodic distortion of the base material 9 by ink is estimated using the estimation model M. In particular, in this printing apparatus 1, the distortion detection unit 60 can remove the periodic distortion component that depends on the position in the conveyance direction and obtain the learning data D8 that extracts only the non-periodic distortion component that depends on the image of the input data. That is, in the learning process of the estimation model M, the accurate correspondence between the input data and the non-periodic distortion of the base material 9 can be learned. And in this printing apparatus 1, in consideration of the estimation result by the estimation model M, ink droplets can be ejected onto the printing surface of the base material 9 while correcting the ejection position of the ink. As a result, a high-quality printed matter with little distortion and little misregistration of the printed image can be obtained.

[0101] <6. Modified Example> As described above, one embodiment of the present invention has been described, but the present invention is not limited to the above-described embodiment.

[0102] <6-1. First Modified Example> In the above-described embodiment, in the learning data D8, a plurality of grid marks 90 were arranged at equal intervals. However, the intervals between the plurality of grid marks 90 do not necessarily have to be equal. For example, the grid marks 90 may be arranged more densely in a portion where the change in the density value of the pattern 91 or the change in the printing rate is large than in other portions.

[0103] <6-2. Second Modified Example> In the above embodiment, the grid mark 90 was composed of a white base figure 901 and a black mark figure 902. However, the color of the base figure 901 is not necessarily limited to white. Also, the color of the mark figure 902 is not necessarily limited to black. For example, the base figure 901 may be black and the mark figure 902 may be white. Further, the base figure 901 and the mark figure 902 may be other colors. That is, for the grid mark 90, it is sufficient that the base figure 901 is formed in a first color and the mark figure 902 is formed in a second color different from the first color. Also, the shapes of the base figure 901 and the mark figure 902 may be different from those in the above embodiment.

[0104] <6-3. Third Modification Example> In the above embodiment, the distortion of the base material 9 was estimated based on the estimation model M generated by machine learning. However, the distortion of the base material 9 may be estimated by other methods. For example, the relationship between the density value or printing rate of each region included in the input data and the distortion of the base material 9 may be formulated. Also, the correspondence between the density value or printing rate of each region included in the input data and the distortion of the base material 9 may be defined by a table. Then, before printing the print input data D3, an estimation result indicating the distortion of the base material 9 may be output based on the print input data D3 and the above formula or table.

[0105] However, when using a formula or a table, every time the state of the printing apparatus 1 or the type of the base material 9 is changed, it is necessary to correct the formula or the table. Since the work of correcting this formula or table needs to be considered from various viewpoints, the burden on the user is large. In contrast, if machine learning is used as in the above embodiment, even when the state of the printing apparatus 1 or the type of the base material 9 is changed, the estimation model M can be recreated by performing a certain learning process. Therefore, it is possible to easily cope with changes in conditions.

[0106] <6-4. Other Modification Examples> Also, in the above embodiment, as shown in FIG. 2, in each of the heads 21 to 24, the nozzles 201 were arranged in a row in the width direction. However, in each of the heads 21 to 24, the nozzles 201 may be arranged in two or more rows.

[0107] Further, the printing apparatus 1 of the above embodiment included four heads 21 to 24. However, the number of heads included in the printing apparatus 1 may be one to three, or five or more. For example, the printing apparatus 1 may include a head that discharges special ink in addition to the inks of each color of K, C, M, and Y.

[0108] Also, the respective elements that appeared in the above embodiment and the modified examples may be appropriately combined within a range where no contradiction occurs.

Explanation of Reference Numerals

[0109] 1 Printing apparatus 9 Base material 40 Camera 51 Input data acquisition unit 52 Distortion measurement unit 53 Data correction unit 54 Learning unit 55 Estimation unit 56 Correction value calculation unit 57 Operation control unit 60 Distortion detection unit 61 First distortion data acquisition unit 61 Second distortion data acquisition unit 62 Second distortion data acquisition unit 63 Shift amount calculation unit 64 Difference processing unit 90 Grid mark 901 Base figure 902 Mark figure D1 Reference input data D2 Evaluation input data D3 Printing input data D6 First distortion data D7 Second distortion data D8 Learning data D9 Estimation result DS shift amount M estimation model

Claims

1. In a printing apparatus that discharges ink onto the surface of a long strip-shaped base material while transporting the base material in the longitudinal direction, a method for detecting two-dimensional distortion of the base material, comprising: a) a first distortion data acquisition step of acquiring first distortion data indicating the distortion of the base material when reference input data is printed; b) a second distortion data acquisition step of acquiring second distortion data indicating the distortion of the base material when evaluation input data is printed; c) a shift amount calculation step of comparing the first distortion data and the second distortion data while shifting them in the longitudinal direction, and calculating a shift amount at which the distortion is most synchronized; d) a data correction step of obtaining correction data by taking the difference between the second distortion data and the first distortion data shifted based on the shift amount. A detection method comprising the above steps.

2. The detection method according to claim 1, wherein each of the steps a) and b) includes: p1) a step of printing an image on the base material based on the input data; p2) a step of imaging the base material on which printing has been performed in step p1); p3) a step of calculating distortion based on the captured image captured in step p2), wherein in step a), the input data is the reference input data, and in step b), the input data is the evaluation input data.

3. The detection method according to claim 2, wherein the reference input data is image data including only a plurality of marks arranged at predetermined intervals, and the evaluation input data includes a plurality of the marks arranged at the same intervals as the reference input data, and a pattern or character for evaluation, and in step p3), distortion is calculated based on the positions of the marks on the base material in the captured image.

4. The detection method according to claim 3, wherein the mark includes a base figure formed in a first color, and a mark figure formed in a second color different from the first color at a position overlapping the base figure.

5. The detection method according to claim 4, wherein the first color is white that does not require ink discharge, and the second color is black.

6. The detection method according to any one of claims 3 to 5, wherein in step p3), distortion is calculated based on a change in the interval between adjacent marks.

7. ​ The detection method according to any one of claims 3 to 5, wherein: In the step p3), strain is calculated based on the amount of displacement of the mark. **Claim 8** The detection method according to any one of claims 3 to 7, wherein: The pattern or character for evaluation includes a plurality of regions having different density values or printing rates. **Claim 9** A learning method for generating an estimation model capable of estimating the strain of the base material by machine learning, comprising: x1) Each step of the detection method according to any one of claims 1 to 8; x2) A learning step of generating an estimation model capable of estimating the strain of the base material by machine learning, using the input manuscript data for evaluation as input variables and the correction data as teacher data; A learning method including the above. **Claim 10** In a printing apparatus that discharges ink onto the surface of a long strip-shaped base material while conveying the base material in the longitudinal direction, an estimation method for estimating the strain of the base material, comprising: y1) Each step of the learning method according to claim 9; y2) A data acquisition step of acquiring printing input manuscript data, which is image data to be printed; y3) An estimation step of outputting an estimation result indicating the strain of the base material due to ink based on the printing input manuscript data before printing the printing input manuscript data; Including: In the step y3), the printing input manuscript data is input to the estimation model generated in the step y1), and the strain output from the estimation model is used as the estimation result. **Claim 11** In a printing apparatus that discharges ink onto the surface of a long strip-shaped base material while conveying the base material in the longitudinal direction, a printing method for printing an image on the base material, comprising: z1) Each step of the estimation method according to claim 10; z2) A printing step of discharging ink onto the surface of the base material while correcting the ink discharge position on the base material based on the estimation result. A printing method that executes the above. **Claim 12** In a printing apparatus that discharges ink onto the surface of a long strip-shaped base material while conveying the base material in the longitudinal direction, a two-dimensional strain detection apparatus for the base material, comprising: A first strain data acquisition unit that acquires first strain data indicating the strain of the base material with respect to reference input manuscript data; A second strain data acquisition unit that acquires second strain data indicating the strain of the base material with respect to evaluation input manuscript data; A shift amount calculation unit that compares the first strain data and the second strain data while shifting them in the longitudinal direction and calculates a shift amount at which the strains are most synchronized. A data correction unit that obtains correction data by taking the difference between the second distortion data and the first distortion data shifted based on the shift amount; A detection device having the above.

Images