Printed image matching method, device, system, storage medium and program product

Abstract

A printed image matching method, device, system, storage medium and program product, comprising: obtaining a large-format electronic image comprising a single-mode electronic image; layering the single-mode electronic image to obtain a finishing layer electronic image and a base layer electronic image; generating base layer printing instructions based on the base layer image, and the position information and shape information of the single-mode electronic image on the large-format electronic image; scanning the surface of a printed product printed based on the base layer printing instructions to obtain a large-format base scanning image; obtaining a base scanning image matched with the base layer electronic image in the large-format base scanning image through a machine learning method; obtaining a finishing instruction image comprising position information and deformation correction based on the base scanning image mapping; generating a large-format finishing instruction image and outputting a finishing layer printing instruction. The present disclosure can improve the quality of printed products in the surface finishing process printing, and solve the problem of large overlay error between the surface finishing process printing layer and the base layer pattern.

Description

Technical Field

[0001] This disclosure relates to the field of printing automation technology, and in particular to a method, apparatus, system, storage medium, and program product for matching printed images in a surface finishing process. Background Technology

[0002] Surface finishing processes for printed materials are a general term for processing techniques that enhance the decorative and functional properties of printed materials through physical or chemical treatments. Their core function is to enhance the visual appeal, tactile texture, and durability of products. Common surface finishing processes include varnishing, lamination, screen printing, UV printing, hot stamping, embossing, and texturing. In printing processes that include surface finishing, the printing pattern may involve one or more processes, and different patterns often use different printing equipment. Therefore, multiple overprinting operations are often required, i.e., printing different patterns in layers. This necessitates precise alignment of images at each process stage; otherwise, misalignment between images printed using different processes will severely affect the quality of the printed product. Printed materials using surface finishing processes are mostly high-end printed products, such as those using hot stamping and screen printing, which have extremely strict requirements for positional accuracy and appearance quality.

[0003] However, surface finishing processes for printed materials involve a variety of materials, including paper, plastic, metal, glass, and fabric. During the printing process, these materials are susceptible to displacement and deformation due to external factors such as material stretching, paper swelling, hot stamping misalignment, and printing errors. This is especially true in large-format printing, where large-format printing presses print multiple pages or large-format prints simultaneously, leading to potentially significant cumulative displacement and deformation. If the image on the print has already become misaligned, deformed, or stretched, and subsequent processes continue printing according to pre-set positions and shapes, the images printed in each process will inevitably misalign, resulting in misalignment between the printed images.

[0004] In existing technologies, the above problems are typically addressed by manual measurement followed by correction. This involves a technician manually measuring the positional deviation and distortion on the image printed in the previous process. Pre-press personnel then adjust the position and compensate for distortion in the subsequent print based on these manual measurements. However, manual measurement has significant errors, and such adjustments often require multiple iterations, resulting in low production efficiency and a waste of raw materials. Furthermore, the printing accuracy after multiple adjustments still may not fully meet quality requirements. Therefore, a new method for matching printed images is needed to solve these problems.

[0005] The above information is presented as background information only to aid in understanding this disclosure. No confirmation or other representation is made regarding whether any of the above constitutes an application of prior art to this disclosure. Summary of the Invention

[0006] The embodiments of this disclosure solve the problem in the aforementioned printing surface finishing process where displacement and deformation occur in the preceding process due to external influencing factors such as material stretching, paper swelling, hot stamping position offset, and printing errors. This results in deviations in shape and position between the image printed in the subsequent process and the image printed in the preceding process, making proper registration and misalignment impossible. A printing image matching method with high response speed, accuracy, and reliability is provided. A printing image matching device, a printing image matching system, a non-temporary storage medium, and a computer program product are also provided to solve the execution problem of the printing image matching method.

[0007] The first aspect of this disclosure provides a printed image matching method, comprising:

[0008] S1: Acquire a large-format electronic image, wherein the large-format electronic image includes a single-mode electronic image;

[0009] S2: The single-mode electronic image is layered to obtain a finishing layer electronic image and a base layer electronic image. The positions and shapes of the finishing layer electronic image and the base layer electronic image have a corresponding relationship in the single-mode electronic image.

[0010] S3: Based on the base layer electronic image, and the position and shape information of the single-mode electronic image on the large-format electronic image, generate base layer printing instructions;

[0011] S4: Scan the surface of the printed material based on the base layer printing instructions to obtain a large-format base scan image;

[0012] S5: Using machine learning methods, obtain a base scan image that matches the base layer electronic image from the large-format base scan image;

[0013] S6: Based on the base scan image and the correspondence between the finishing layer electronic image and the base layer electronic image, a finishing instruction image including position information and deformation correction is obtained by mapping.

[0014] S7: Generate a large-format finishing instruction image based on the finishing instruction image, and output finishing layer printing instructions based on the large-format finishing instruction image.

[0015] For example, in at least one embodiment, it further includes:

[0016] S8: Evaluate the matching effect between the base scan image and the finishing instruction image based on the reinforcement learning algorithm, and optimize the model based on the evaluation results;

[0017] The core formula of the reinforcement learning algorithm is:

[0018]

[0019] Where s is the state, a is the action, Q(s,a) is the expected return of taking action a in state s, r is the immediate return, γ is the discount rate, s′ is the next state, and a′ is the next action.

[0020] For example, in at least one embodiment, image parameters, positional error values, and shape error values ​​of the base scan image and the finishing instruction image are calculated, and a machine adjustment threshold is set. When the image parameters, positional error values, or shape error values ​​exceed the machine adjustment threshold, preset parameters of the surface finishing process printing machine are adjusted based on preset feedback control logic. The preset parameters include one or more of tension, air pressure, and printing speed.

[0021] For example, in at least one embodiment, the large-format electronic image in step S1 includes multiple single-mode electronic images, wherein each single-mode electronic image includes a finishing layer electronic image and a base layer electronic image; in step S5, multiple base scan images matching the base layer electronic image are obtained; in step S6, multiple finishing instruction images are obtained based on the multiple base scan images and the correspondence between the finishing layer electronic image and the base layer electronic image in the single-mode electronic image; in step S7, the large-format finishing instruction image is generated based on the multiple finishing instruction images and their position and shape information.

[0022] For example, in at least one embodiment, the scanning of the printed surface based on the base layer printing instructions in step S4 is performed in real time during the finishing layer printing process; the large-format base scan image is divided into h rows along the printing direction, and step S5 is performed once for each row after printing begins to obtain the base scan image in the subsequent rows.

[0023] For example, in at least one embodiment, step S2, which involves layering the single-mode electronic image, further includes obtaining a die-cutting layer, the die-cutting layer including die-cutting lines and position and shape information of the die-cutting lines; before step S5, the step further includes determining the image matching region in the large-format basic scan image based on the die-cutting lines.

[0024] For example, in at least one embodiment, the machine learning algorithm in step S5 is a convolutional neural network algorithm, and its core convolutional layer algorithm is expressed as follows:

[0025]

[0026] Among them, F i,j It is a pixel in the feature map, I x,y W is the number of pixels in the input image. x,y These are the filter weights;

[0027] The machine learning method in step S5 is a support vector machine (SVM), and its core algorithm is expressed as:

[0028]

[0029] where f(x) is the function value to be predicted, αi is the Lagrange multiplier, yi is the class label of the sample, K is the kernel function, and b is the bias.

[0030] For example, in at least one embodiment, the base layer electronic image is divided into n layers and printed in n times; wherein, during the m-th (1 < m <= n) printing, the surface of the printed matter with the 1st to (m - 1)-th base layer electronic images is scanned to obtain the (m - 1)-th large-format base scan image; through the machine learning method, in the (m - 1)-th large-format base scan image, the (m - 1)-th base scan image matching the (m - 1)-th base layer electronic image is obtained; or, the finishing layer electronic image is divided into k layers, and k surface finishing process printing layers are printed by repeating steps S3 - S7 k times.

[0031] The second aspect of the present disclosure provides a printing image matching device, including: a layout module, a layering module, a base layer printing module, a scanning module, an acquisition module, an instruction module, and an evaluation module; wherein, the layout module is configured to obtain a large-format electronic image, and the large-format electronic image includes a single-mode electronic image; the layering module is configured to layer the single-mode electronic image to obtain a finishing layer electronic image and a base layer electronic image, and there is a corresponding relationship between the positions and shapes of the finishing layer electronic image and the base layer electronic image in the single-mode electronic image; the base layer printing module is configured to generate a base layer printing instruction based on the base layer electronic image, as well as the position information and shape information of the single-mode electronic image on the large-format electronic image; the scanning module is configured to scan the surface of the printed matter printed based on the base layer printing instruction to obtain a large-format base scan image; the matching module is configured to obtain a base scan image matching the base layer electronic image in the large-format base scan image through a machine learning method; the acquisition module is configured to map and obtain a finishing instruction image including position information and deformation correction based on the base scan image and the corresponding relationship between the finishing layer electronic image and the base layer electronic image; the instruction module is configured to generate a large-format finishing instruction image based on the finishing instruction image and output a finishing layer printing instruction based on the large-format finishing instruction image; the evaluation module is configured to evaluate the matching effect of the base scan image and the finishing instruction image based on a reinforcement learning algorithm, and optimize the model according to the evaluation result.

[0032] A third aspect of this disclosure provides a printed image matching system, comprising: a memory for non-temporarily storing computer-executable instructions; and a processor for running the computer-executable instructions, wherein the computer-executable instructions, when run by the processor, perform the printed image matching method described in any of the preceding claims.

[0033] A fourth aspect of this disclosure provides a non-transitory storage medium for storing computer-executable instructions, wherein, when the computer-executable instructions are executed by a computer, the printed image matching method described in any of the preceding claims is performed.

[0034] The fifth aspect of this disclosure provides a computer program product, including a computer program, characterized in that, when executed by a processor, the computer program implements the printed image matching method described in any of the preceding claims. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. Obviously, the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure.

[0036] Figure 1 This is a flowchart of a printing image matching method according to an embodiment of the present disclosure;

[0037] Figure 2 This is a schematic diagram of a first large-format electronic image according to an embodiment of the present disclosure;

[0038] Figure 3 This is a second large-format electronic image schematic diagram according to an embodiment of the present disclosure;

[0039] Figure 4 It is a single-mode electronic image layering map according to an embodiment of the present disclosure;

[0040] Figure 5 This is a large-format basic image schematic diagram according to an embodiment of the present disclosure;

[0041] Figure 6 This is a schematic diagram of a large-format basic scanned image according to an embodiment of the present disclosure;

[0042] Figure 7 This is a schematic diagram of basic scan image matching according to an embodiment of the present disclosure;

[0043] Figure 8 This is a schematic diagram of finishing instructions according to an embodiment of the present disclosure;

[0044] Figure 9 This is a scanned schematic diagram after printing according to an embodiment of the present disclosure;

[0045] Figure 10 This is a schematic diagram of online scanning according to an embodiment of the present disclosure;

[0046] Figure 11 This is a schematic diagram of the basic layer electronic image layering according to an embodiment of the present disclosure;

[0047] Figure 12 This is a schematic diagram of a third large-format electronic image according to an embodiment of this disclosure;

[0048] Figure 13 This is a schematic diagram of multiple matching of a basic scan image according to an embodiment of the present disclosure. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0050] Unless otherwise defined, the technical or scientific terms used in this disclosure shall have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms “first,” “second,” and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as “comprising” or “including” mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as “connected” or “linked” are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as “upper,” “lower,” “left,” and “right” are used only to indicate relative positional relationships, and these relative positional relationships may change accordingly when the absolute position of the described objects changes. In this disclosure, “multiple” means two or more.

[0051] According to embodiments of this disclosure, Figure 1 A method for matching printed images is disclosed, comprising:

[0052] S1: Obtain large-format electronic image 1, which includes single-mode electronic image 11. Large-format electronic image 1 is a design pattern for the printing task and is the electronic image that serves as the final target for large-format printing. Figure 2 As shown, the large-format electronic image 1 can be a large-format single-mode image adapted for printing on a large-format printing press. In this case, the large-format electronic image 1 includes a single-mode electronic image 11. For example... Figure 3The large-format electronic image 1 shown can also be a composite image obtained through printing imposition technology. In this case, the large-format electronic image 1 includes a combination of multiple single-mode electronic images 11. The large-format electronic image 1 includes not only the image information of each single-mode electronic image 11, but also the position and shape information of each single-mode electronic image 11 within the large-format electronic image 1. Typically, the position and shape information of each single-mode electronic image 11 within the large-format electronic image 1 can be displayed independently in the software. The large-format electronic image 1 is not only the foundational image for subsequent printing processes, but also the standard for quality inspection and continuous optimization after the printed product is completed.

[0053] S2: As Figure 4 As shown, the single-mode electronic image 11 is layered to obtain a finishing layer electronic image 112 and a base layer electronic image 111. The positions and shapes of the finishing layer electronic image 112 and the base layer electronic image 111 correspond to each other in the single-mode electronic image 11. In some embodiments, layering the single-mode electronic image 11 further includes obtaining a die-cutting line layer. The die-cutting lines are not displayed in the printed material and are process lines reserved in the software. Before actual printing, according to the printing process, the image for each printing is divided into one layer on the single-mode electronic image 11. For example, in Figure 4 In this embodiment, the image printed first is classified as a base layer electronic image 111, and the image printed using a surface finishing process is classified as a finishing layer electronic image 112. Depending on the complexity of the printed matter and the choice of printing process, the base layer electronic image 111 may require multiple printings, and therefore may include multiple layers. The finishing layer electronic image 112 may include various surface finishing process steps, and may also include multiple layers.

[0054] S3: Based on the base layer electronic image 111 on the large-format electronic image 1, and the position and shape information of the single-mode electronic image 11 on the large-format electronic image 1, a base layer printing instruction is generated. After the single-mode electronic image 11 is layered, the base layer electronic image 111 is used as the printing image of the base layer. For example... Figure 5 Based on the position and shape information of each single-mode electronic image 11 on the large-format electronic image 1, a large-format base image 2 corresponding to the base layer electronic image 111 is generated. In the large-format base image 2, the base layer printing image 211 is set at the corresponding position of each single-mode electronic image 11. Based on the large-format base image 2, a base layer printing instruction is generated. At this time, the base layer printing instruction includes the color and graphic information of the document to be printed. After executing the base layer printing instruction, the following can be obtained: Figure 5 The pattern and its colors are shown. The large-format printing press executes the base layer printing instructions to complete the printing of the base layer image.

[0055] S4: As Figure 6As shown, the surface of the printed material printed based on the base layer printing instructions is scanned to obtain a large-format base scan image 3. This large-format base scan image 3 can be scanned offline or online. Offline scanning refers to transferring the printed material with the base layer printed on it to a scanner after the base layer printing is completed. Online scanning refers to setting up a scanner on a surface finishing printing machine, fixing the surface finishing printed material on the machine, and then scanning it online. Scanning can be performed offline, but for materials prone to process deformation such as fabrics, plastics, and paper, online scanning is preferred to eliminate errors caused by fixing the printing material and to meet the needs of real-time scanning and dynamic adjustment.

[0056] S5: As Figure 7 As shown, a base scan image 311 matching the base layer electronic image 111 is obtained from the large-format base scan image 3 using machine learning methods. Specifically, the image and contour features of the base layer electronic image 111 are learned using machine learning methods, and base scan images 311 with similar image and contour features are matched from the large-format base scan image 3. Each matched base scan image 311 includes its positional information and deformation information. The large-format base scan image 3 may also include the matched scanning die-cutting line 313.

[0057] S6: As Figure 8 As shown, based on the base scan image 311 and the correspondence between the finishing layer electronic image 112 and the base layer electronic image 111, a finishing instruction image 412 including position information and deformation correction is obtained by mapping. After matching the base scan image 311 on the large-format base scan image 3, the position information and deformation information of the base scan image 311 are mapped to the corresponding surface finishing process image, and the optimal matching effect can be calculated. After performing deformation correction on each finishing layer electronic image 112 on the large-format electronic image 1, the printing shape of each finishing instruction image 412 is obtained. According to the position information of the corresponding base scan image 311, each finishing instruction image 412 is assigned corresponding position information.

[0058] S7: Generate a large-format finishing instruction image 4 based on the finishing instruction image 412, and output a finishing layer printing instruction based on the large-format finishing instruction image 4. The finishing layer printing instruction includes processing information for the areas requiring surface finishing. After executing the finishing layer printing instruction, a printed product including the surface finishing process can be obtained. In step S6, each deformed and corrected finishing instruction image 412 and its position information have been obtained. All finishing instruction images 412 are set in the large-format according to the position correction information, thus obtaining the large-format finishing instruction image 4. Instructions are sent to the surface finishing process printing machine based on the large-format finishing instruction image 4. The surface finishing process printing machine completes the printing.

[0059] Preferably, the printed image matching method further includes S8: evaluating the matching effect between the base scan image 311 and the finishing instruction image 412 based on a reinforcement learning algorithm, and optimizing the model based on the evaluation results. The base scan image 311 is an electronic image obtained from scanning, and the finishing instruction image 412 is an electronic image obtained after image processing. Figure 9 As shown, after printing the finishing instruction image 412 on the printed material, the entire printed material can be scanned to obtain the scanned image 5. From the scanned image 5, one or more single-mode actual images 51 corresponding to the single-mode electronic image 11 can be extracted. The single-mode actual image 51 includes the base layer actual image 511 and the finishing actual image 512, and their relative positions and shape relationships are extracted. Theoretically, ideally, the relative positions and shapes of each pair of base layer actual images 511 and finishing actual images 512 should be exactly the same as the relative positions and shapes of the base layer electronic image 111 and the finishing layer electronic image 112. However, in actual printing processes, there will inevitably be accumulated errors from various factors. Large errors mean poor matching results, while small errors mean good matching results.

[0060] Reinforcement learning algorithms collect multiple error parameters as states and use each match as an action to optimize the machine learning model.

[0061] The core formula of reinforcement learning algorithms is:

[0062]

[0063] Where s is the state, a is the action, Q(s,a) is the expected return of taking action a in state s, r is the immediate return, γ is the discount rate, s′ is the next state, and a′ is the next action.

[0064] Preferably, the image parameters, positional error values, and shape error values ​​of the base scan image 311 and the finishing instruction image 412 are calculated, and a machine adjustment threshold is set. When the calculated image parameters, positional error values, or shape error values ​​exceed the machine adjustment threshold, the preset parameters of the surface finishing process printing press are adjusted based on preset feedback control logic. The preset parameters include one or more of tension, air pressure, and printing speed. For example, when the shape error of the base scan image 311 exceeds the shape error threshold, the machine learning model can output control information to the printing press feedback control module to reduce the printing press tension. For example, when a defect with poor printing fill is detected, the machine learning model can output control information to the printing press feedback control module to reduce the printing speed. Printing equipment is subject to various interference factors during operation, such as changes in ambient temperature and humidity, mechanical vibration, and electrical system noise. These interference factors can cause the printing press parameter settings to fail to meet printing requirements, reducing printing quality. Based on the image evaluation by the machine learning module, the machine parameters of the surface finishing process printing press can be ensured to always be in an optimal state.

[0065] Preferred, such as Figure 3 As shown, in step S1, the large-format electronic image 1 includes multiple identical single-mode electronic images 11, wherein each single-mode electronic image 11 includes a finishing layer electronic image 112 and a base layer electronic image 111. For example... Figure 7 As shown, in step S5, multiple basic scan images 31 matching the basic layer electronic image are acquired. For example... Figure 8 As shown, in step S6, multiple finishing instruction images 412 are obtained based on multiple basic scan images 31 and the correspondence between the basic layer electronic image 111 and the finishing layer electronic image 112 in the single-mode electronic image 11. In step S7, a large-format finishing instruction image 4 is generated based on the multiple finishing instruction images 412 and their position and shape information, and the large-format finishing instruction image 4 is printed with a surface finishing process.

[0066] Preferably, the scanning of the printed surface based on the base layer printing instructions in step S4 is performed in real time during the printing process. Figure 10 The diagram shows the structure of a printing press for online inspection, including a scanning device 1001, a vacuum-sealed chamber 1002, a display computer 1003, a keypad 1004, a drawer 1005, a front door 1006, a side panel 1007, and a bottom support 1008. Figure 10 In this process, the printed matter is vacuum-adsorbed onto the machine surface, and the scanning device 1001 performs an online scanning task. The scanning result can be displayed in real time on the display computer 1003 and can be transmitted to the background program for processing in the embodiments of this disclosure.

[0067] In a preferred embodiment of online real-time scanning, the large-format base scan image 3 is divided into h rows along the printing direction. After printing begins, step S5 is executed once for each row printed to obtain the base scan image 311 for the subsequent rows, and the large-format finishing instruction image 4 is updated in real time. For example, in Figure 6 In this embodiment, the surface of the printed material is divided into three rows along the printing direction according to the array rules of the single-mode electronic image 11. After the first row is printed, the base scan image 311 of the second row is scanned. At this time, the position and deformation information of the base scan image 311 has eliminated the errors caused by the printing process of the first row. Similarly, after the second row is printed, the base scan image 311 of the third row is scanned to eliminate the errors caused by the printing processes of the first and second rows to the third row.

[0068] Preferably, in S4, the surface of the printed matter printed based on the base layer printing instructions is scanned, and the large-format base scan image 3 is restored according to a 1:1 ratio between the physical object and the electronic image.

[0069] Preferably, step S2, which involves layering the single-mode electronic image 11, also includes acquiring a die-cutting layer 113. This die-cutting layer includes die-cutting lines and their position and shape information. Before step S5, a step is also included to determine the image matching area in the large-format base scan image 3 based on the die-cutting lines. In the packaging printing field, die-cutting lines are a technical drawing showing how to print and assemble packaging information, and are typically not printed on the printed material. Die-cutting lines usually start with basic packaging information, such as height, width, and depth. It also includes the dimensions of each side panel and flap, as well as any cuts or holes that need to be created. Therefore, die-cutting lines can accurately determine the final product printing area; the area outside the die-cutting lines is the area that will ultimately be cut off and does not require much attention. Large-format printed materials have large formats; if machine learning is performed on the image within the entire large-format area, the computational resources required by the algorithm when processing the data would be too large, leading to low machine learning computational efficiency. Before step S5, the position and shape of the die-cutting line in the large-format base scan image 3 are first confirmed. Then, machine learning is performed only within the die-cutting line region of the large-format base scan image 3. Within each die-cutting line region of the large-format base scan image 3, the base scan image 311 is matched. The image area within the die-cutting line region is much smaller than the area of ​​the entire large-format image, thus effectively saving computational resources and improving the efficiency of machine learning.

[0070] Preferably, the position and shape of the die-cutting line in the large-format base scan image 3 are determined firstly based on the position and shape information of the die-cutting line in the die-cutting layer 113, and the position and shape information of the single-mode electronic image 11 containing the die-cutting layer 113 in the large-format electronic image 1. The position and shape information of the die-cutting line in the large-format electronic image 1 is directly used as the position and shape information of the die-cutting line in the large-format base scan image 3.

[0071] Preferably, to determine the position and shape of the die-cutting line in the large-format base scan image 3, the method in step S5 is first used to obtain the edge contours of all images in the large-format base scan image 3 through machine learning, and the die-cutting line is obtained by matching the edge contours.

[0072] Preferably, in step S6, the coordinate difference and deformation ratio of the finishing layer electronic image 112 and the base layer electronic image 111 are calculated. The coordinate values ​​of the base scan image 311 on the large-format base scan image 3 are added to the calculated coordinate difference to obtain the coordinate values ​​of the corresponding finishing instruction image 412. The deformation information of the base scan image 311, such as the relative positions of feature points, is multiplied by the calculated deformation ratio to obtain the shape information of the finishing instruction image 412. Directly determining the position and shape of the finishing instruction image 412 based on the base scan image 311 allows for optimal matching between the two, which is beneficial for improving the quality of the single-mode image.

[0073] Preferably, in step S6, the base layer electronic images 111 are assembled into a larger version based on the large-format electronic image 1. The large-format image of the base layer electronic image 111 is aligned with the large-format base scan image 3, that is, the large-format image of the base layer electronic image 111 is given a positional offset and deformation according to the overall printing situation of the large-format base scan image 3. Then, as needed, secondary corrections are performed based on the deformations of the multiple base scan images 311 and their relative positions to determine the position and shape of the finishing instruction image 412. This algorithm helps ensure consistency between individual models on the printed material.

[0074] The coordinate difference and deformation ratio of the finishing layer electronic image 112 and the base layer electronic image 111 are calculated. The coordinate values ​​of the base scan image 311 on the large-format base scan image 3 are added to the calculated coordinate difference to obtain the coordinate values ​​of the corresponding finishing instruction image 412. The deformation information of the base scan image 311, such as the relative positions of feature points, is multiplied by the calculated deformation ratio to obtain the shape information of the finishing instruction image 412. Determining the position and shape of the finishing instruction image 412 directly based on the base scan image 311 achieves the best match between the two, which is beneficial for improving the quality of the single-mode image.

[0075] Preferably, the machine learning algorithm in S5 is a convolutional neural network algorithm, and its convolutional layer is represented as follows:

[0076]

[0077] Among them, F i,j It is a pixel in the feature map, I x,y W is the number of pixels in the input image. x,y These are the filter weights;

[0078] The machine learning method of S5 is the support vector machine (SVM), and its core algorithm is expressed as:

[0079]

[0080] where f(x) is the function value to be predicted, αi is the Lagrange multiplier, yi is the class label of the sample, K is the kernel function, and b is the bias.

[0081] Preferably, the convolutional neural network model includes an input layer, a convolutional layer, a ReLU layer, a pooling layer, and a fully connected layer. The foregoing layers are stacked to construct a complete convolutional neural network. The convolutional layer is the core layer of constructing the convolutional neural network,承担主要的计算量,通过过滤器提取图像的特征。ReLU层基于ReLU激活函数优化图像特征。在计算量过大时,训练参数被要求在随后的卷积层之间周期性地引进池化层,以减少图像的空间大小。全连接层中,下一层的每一个神经元与上一层的神经元全部相连,以实现图像特征的输出。

[0082] Preferably, in step S7, in the large-format finishing instruction image 4 generated based on the finishing instruction image 412, it not only includes the information such as the image contour, color, position information, deformation information, etc. of one or more finishing instruction images 412 respectively, which are reflected on the final printed product, but also includes process information such as control strips, positioning lines, color references, etc., to ensure the perfect process implementation of the surface finishing process printing layer.

[0083] Preferably, as Figure 11 shown, the base layer electronic image 111 is divided into n layers and printed in n times. Among them, when printing for the mth (1 < m <= n) time, the surface of the printed product with the 1st to (m - 1)th base layer electronic images is scanned to obtain the (m - 1)th large-format base scan image; through the machine learning method, in the (m - 1)th large-format base scan image, the (m - 1)th base scan image matching the (m - 1)th base layer electronic image is obtained. For example, in Figure 11 the base layer electronic image 111 is divided into 2 layers, including the 1st base layer electronic image 1111 and the 2nd base layer electronic image 1112. When printing the 1st layer, it is printed based on the position and shape of the single-mode electronic image 11 in the large-format electronic image 1. When printing the 2nd layer, the surface of the printed product with the 1st base layer electronic image 1111 is scanned to obtain the 1st layer's large-format base scan image 3. Through the machine learning method, in the (m - 1)th base layer's large-format base scan image 3, the base scan image 311 matching the (m - 1)th base layer's base layer electronic image is obtained.

[0084] Preferably, the finishing layer electronic image 112 in S2 is divided into k layers, and the printing of k surface finishing process printing layers is completed by repeating steps S3-S7 k times. It is understood that the finishing layer electronic image 112 can also be divided into multiple layers, and optimal matching between layers is achieved by repeating steps S3-S7 multiple times. The k-layer finishing layer electronic image 112 may include different surface finishing processes; for example, the first layer is an embossing layer, the second layer is a screen printing layer, and the third layer is a hot stamping layer.

[0085] Preferably, the large-format electronic image 1 includes multiple single-mode electronic images. For example... Figure 12 As shown, the large-format electronic image 1 includes a single-mode electronic image 11 and a second single-mode electronic image 12, both of which include surface finishing printing patterns. In step S2, these are layered. The finishing layer electronic image 112 and the base layer electronic image 111 are obtained, as are the second finishing layer electronic image 122 and the second base layer electronic image 121. In step S3, the base layer electronic image 111 and the second base layer electronic image 121 are combined to obtain the large-format base image 2. Step S5 is executed multiple times according to the type of single-mode electronic image, for example... Figure 13 As shown, the first machine learning process acquires a base scan image 311 that matches the base layer electronic image 111, which may also include die-cutting lines 313; the second machine learning process acquires a second base scan image 321 that matches the second base layer electronic image 121, which may also include die-cutting lines 323. In step S6, the position and shape information of the corresponding base scan images 311 and 321 are used to assign corresponding position and shape information to the finishing instruction image 412 and the finishing instruction image 422. In step S7, the finishing instruction image 412 and the finishing instruction image 422 are combined to obtain a large-format finishing instruction image 4, and printing instructions are output based on the large-format finishing instruction image 4.

[0086] According to an embodiment of this disclosure, a printed image matching device is also provided, comprising: an imposition module, a layering module, a base layer printing module, a scanning module, an acquisition module, an instruction module, and an evaluation module; wherein, the imposition module is configured to acquire a large-format electronic image 1, the large-format electronic image 1 including a single-mode electronic image 11; the layering module is configured to layer the single-mode electronic image 11 to acquire a finishing layer electronic image 112 and a base layer electronic image 111, the positions and shapes of the finishing layer electronic image 112 and the base layer electronic image 111 corresponding to each other in the single-mode electronic image 11; the base layer printing module is configured to generate a base layer printing instruction based on the base layer electronic image 111 and the position and shape information of the single-mode electronic image 11 on the large-format electronic image 1; the scanning module is configured to scan the base layer electronic image 111 and the single-mode electronic image 11 on the large-format electronic image 1. The surface of the printed material printed by the layer printing instruction is used to obtain a large-format basic scan image 3. The matching module is configured to use a machine learning method to obtain a basic scan image 311 that matches the basic layer electronic image 111 in the large-format basic scan image 3. The acquisition module is configured to map a finishing instruction image 412, including position information and deformation correction, based on the basic scan image 311 and the correspondence between the finishing layer electronic image 112 and the basic layer electronic image 111. The instruction module is configured to generate a large-format finishing instruction image 4 based on the finishing instruction image 412 and output the finishing layer printing instruction based on the large-format finishing instruction image 4. The evaluation module is configured to evaluate the matching effect of the basic scan image 311 and the finishing instruction image 412 based on the reinforcement learning algorithm and optimize the model according to the evaluation results.

[0087] According to an embodiment of this disclosure, a printed image matching system is also provided, comprising: a memory for non-temporarily storing computer-executable instructions; and a processor for running the computer-executable instructions, wherein the computer-executable instructions are executed by the processor to perform the printed image matching method of any of the above embodiments.

[0088] According to embodiments of this disclosure, a non-transitory storage medium is also provided for storing computer-executable instructions, wherein when the computer-executable instructions are executed by a computer, the printing image matching method of any of the above embodiments is executed.

[0089] According to an embodiment of this disclosure, a computer program product is also provided, including a computer program, characterized in that, when the computer program is executed by a processor, it implements the printing image matching method of any of the above embodiments.

[0090] The following points also need to be explained:

[0091] (1) The accompanying drawings of the embodiments of this disclosure only involve the structures involved in the embodiments of this disclosure, and other structures can be referred to the general design.

[0092] (2) For clarity, the thickness of devices, layers, or regions is enlarged or reduced in the drawings used to describe embodiments of the present disclosure, i.e., these drawings are not drawn to scale. It will be understood that when an element such as a layer, film, region, or substrate is referred to as being “above” or “below” another element, the element may be “directly” located “above” or “below” the other element or there may be intermediate elements.

[0093] (3) Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

[0094] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. The scope of protection of this disclosure shall be determined by the scope of the claims.

Claims

1. A method for matching printed images, comprising: S1: Acquire a large-format electronic image, wherein the large-format electronic image includes a single-mode electronic image; S2: The single-mode electronic image is layered to obtain a finishing layer electronic image and a base layer electronic image. The positions and shapes of the finishing layer electronic image and the base layer electronic image have a corresponding relationship in the single-mode electronic image. S3: Based on the base layer electronic image, and the position and shape information of the single-mode electronic image on the large-format electronic image, generate base layer printing instructions; S4: Scan the surface of the printed material based on the base layer printing instructions to obtain a large-format base scan image; S5: Using machine learning methods, a base scan image matching the base layer electronic image is obtained from the large-format base scan image. The machine learning algorithm is a convolutional neural network algorithm, and its core convolutional layer algorithm is expressed as follows: Where Fi,j is a pixel in the feature map, Ix,y is a pixel in the input image, and Wx,y is the filter weight. The machine learning method mentioned is Support Vector Machine (SVM), and its core algorithm is expressed as follows: Where f(x) is the function value to be predicted, αi is the Lagrange multiplier, yi is the class label of the sample, K is the kernel function, and b is the bias; S6: Based on the base scan image and the correspondence between the finishing layer electronic image and the base layer electronic image, a finishing instruction image including position information and deformation correction is obtained by mapping. S7: Generate a large-format finishing instruction image based on the finishing instruction image, and output finishing layer printing instructions based on the large-format finishing instruction image. The large-format finishing instruction image includes the image outline, color, position, control bars, positioning lines, and color reference information of the finishing instruction image.

2. The printed image matching method according to claim 1, characterized in that, Also includes: S8: Evaluate the matching effect between the base scan image and the finishing instruction image based on the reinforcement learning algorithm, and optimize the model based on the evaluation results; The core formula of the reinforcement learning algorithm is: Where s is the state, a is the action, Q(s,a) is the expected return of taking action a in state s, r is the immediate return, γ is the discount rate, s′ is the next state, and a′ is the next action.

3. The printed image matching method according to claim 2, characterized in that, Calculate the image parameters, position error value, and shape error value of the base scan image and the finishing instruction image, and set the machine adjustment threshold; When the image parameter, position error value, or shape error value exceeds the machine adjustment threshold, the preset parameters of the surface finishing process printing machine are adjusted based on the preset feedback control logic. The preset parameters include one or more of tension, air pressure, and printing speed.

4. The printed image matching method according to claim 1, characterized in that, The large-format electronic image mentioned in step S1 includes multiple single-mode electronic images, wherein each single-mode electronic image includes a finishing layer electronic image and a base layer electronic image; In step S5, multiple basic scan images matching the basic layer electronic image are acquired; In step S6, multiple finishing instruction images are obtained based on the multiple base scan images and the correspondence between the finishing layer electronic image and the base layer electronic image in the single-mode electronic image; In step S7, the large-format finishing instruction image is generated based on the multiple finishing instruction images and their position and shape information.

5. The printed image matching method according to claim 4, characterized in that, In step S4, the surface of the printed material is scanned based on the printing instructions of the base layer, which is done in real time during the finishing layer printing process; The large-format basic scan image is divided into h rows along the printing direction. After printing begins, step S5 is executed once for each row printed to obtain the basic scan image for the subsequent rows.

6. The printed image matching method according to claim 1, characterized in that, Step S2 involves layering the single-mode electronic image and also includes obtaining a die-cutting layer, which includes die-cutting lines and their position and shape information. Before step S5, the method further includes a step of determining the image matching region in the large-format base scan image based on the die-cutting line.

7. The printed image matching method according to claim 1, characterized in that, The base layer electronic image is divided into n layers and printed n times; In the m-th printing, the surface of the printed material bearing the electronic images of the first to m-1th base layers is scanned to obtain the m-1th large-format base scan image, where 1 <m<=n; Using machine learning methods, the (m-1)th base scan image that matches the (m-1)th base layer electronic image is obtained from the (m-1)th large-format base scan image; Alternatively, the finishing layer electronic image is divided into k layers, and the printing of k surface finishing process printing layers is completed by repeating steps S3-S7 k times.

8. A printed image matching device, comprising: The module includes imposition, layering, base layer printing, scanning, matching, acquisition, instruction, and evaluation modules; among them, The imposition module is configured to acquire a large-format electronic image, which includes a single-mode electronic image. The layering module is configured to layer the single-mode electronic image to obtain a finishing layer electronic image and a base layer electronic image, wherein the positions and shapes of the finishing layer electronic image and the base layer electronic image correspond to each other in the single-mode electronic image; The base layer printing module is configured to generate base layer printing instructions based on the base layer electronic image, as well as the position and shape information of the single-mode electronic image on the large-format electronic image; The scanning module is configured to scan the surface of printed materials printed based on the base layer printing instructions to obtain a large-format base scan image; The matching module is configured to use a machine learning method to obtain a base scan image that matches the base layer electronic image from the large-format base scan image. The machine learning algorithm is a convolutional neural network algorithm, and its core convolutional layer algorithm is expressed as follows: Where Fi,j is a pixel in the feature map, Ix,y is a pixel in the input image, and Wx,y is the filter weight. The machine learning method mentioned is Support Vector Machine (SVM), and its core algorithm is expressed as follows: Where f(x) is the function value to be predicted, αi is the Lagrange multiplier, yi is the class label of the sample, K is the kernel function, and b is the bias; The acquisition module is configured to map a finishing instruction image, including position information and deformation correction, based on the base scan image and the correspondence between the finishing layer electronic image and the base layer electronic image. The instruction module is configured to generate a large-format finishing instruction image based on the finishing instruction image, and to output finishing layer printing instructions based on the large-format finishing instruction image. The large-format finishing instruction image includes the image outline, color, position, control bars, positioning lines, and color reference information of the finishing instruction image. The evaluation module is configured to evaluate the matching effect between the base scan image and the finishing instruction image based on a reinforcement learning algorithm, and optimize the model based on the evaluation results.

9. A printed image matching system, comprising: Memory is used to store non-temporary executable instructions for a computer. And a processor for running the computer-executable instructions, wherein the computer-executable instructions are executed by the processor to perform the printed image matching method according to any one of claims 1 to 7.

10. A non-transitory storage medium for non-transitory storage of computer-executable instructions, wherein, When the computer-executable instructions are executed by a computer, the printed image matching method according to any one of claims 1 to 7 is performed.

11. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the printed image matching method according to any one of claims 1 to 7.

Images