A printer waveform automatic screening method, device, equipment and medium

Abstract

The application discloses a printer waveform automatic screening method, device, equipment and medium, and relates to the technical field of printer waveform adjustment, to solve the problems that the existing waveform screening needs manual one-by-one evaluation of a printing image, the accuracy has deviation, and the time consumption is long. The method comprises the following steps: acquiring historical printing image data; determining the score evaluation value of each historical printing image by adopting a crowd-sourcing Bradley-Terry model algorithm; constructing a score prediction model initial model; training the score prediction model initial model based on the score evaluation value and the historical printing image, to obtain a score prediction model; feeding a to-be-predicted printing image into the score prediction model, to predict the prediction score corresponding to the to-be-predicted printing image; and sorting the to-be-screened waveforms, and determining the top M to-be-screened waveforms as target waveforms. The printer waveform automatic screening method provided by the application can screen the printer waveform without manual operation, has high waveform screening speed and high accuracy.

Description

Technical Field

[0001] This invention relates to the field of printer waveform adjustment technology, and in particular to a method, apparatus, equipment and medium for automatic waveform selection in printers. Background Technology

[0002] Printer waveform adjustment is a core step in ensuring inkjet printing quality. By optimizing voltage timing, it controls droplet formation and ejection precision, directly impacting output quality. Currently, a batch of waveforms is typically guessed manually, the printer prints images based on these guesses, and then each image is scored manually. Based on the scores, the highest quality waveforms are selected. However, manually scoring images requires a long time to train a professional, and the transmission of experience takes time. Furthermore, subjective evaluation is prone to bias, and scoring all images manually is time-consuming. Summary of the Invention

[0003] The purpose of this invention is to provide a method, apparatus, device, and medium for automatic waveform screening of printers, which solves the problems of existing waveform screening methods that require manual evaluation of each printed image, resulting in inaccuracies, time-consuming processes, and difficulties in training professional evaluators.

[0004] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides an automatic waveform selection method for printers, comprising: Retrieve historical printed image data; Based on the historical printed image data, the crowdsourced Bradley-Terry model algorithm is used to determine the score estimate for each historical printed image; Construct an initial model for the score prediction model; the last layer of the initial model for the score prediction model is a regression head structure. The initial model of the score prediction model is trained based on the score estimate and the historical printed image to obtain the score prediction model; The printed image to be predicted is fed into the score prediction model to predict the predicted score corresponding to the printed image; the printed image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveforms to be screened are sorted according to the predicted scores, and the top M waveforms to be screened are determined as target waveforms, where M is a positive integer greater than 0.

[0005] Optionally, the historical printed image data includes: historical printed images, image numbers, and a comparison result set; the comparison result set includes the number of each group of historical printed images, the winner image number corresponding to each group of historical printed images, and the scorer number; the step of determining the score estimate of each historical printed image using the crowdsourcing Bradley-Terry model algorithm based on the historical printed image data includes: Formula used: Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, For the probability of winning, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; With the goal of maximizing the likelihood function, an optimization algorithm is used to iteratively update the target parameters of the likelihood function until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the rater's credibility, and the rater's preference bias.

[0006] Optionally, the initial model of the score prediction model further includes a backbone network, and the step of training the initial model of the score prediction model based on the score estimate and the historical printed image to obtain the score prediction model includes: The historical printed image is input into the initial model of the score prediction model, and the backbone network extracts morphological features from the historical printed image; the regression head structure processes the morphological features to obtain training prediction values; Calculate the mean squared error loss between the training prediction value and the score estimate, and adjust the parameters of the initial model of the score prediction model according to the mean squared error loss to obtain the score prediction model.

[0007] Optionally, calculating the mean squared error loss between the training prediction and the score estimate includes: Formula used: Calculate the mean squared error loss between the training prediction and the score estimate; in, For mean square error loss, The number of ratings estimated, For rating and valuation, These are the training prediction values.

[0008] Optionally, the step of feeding the print image to be predicted into the score prediction model to predict the predicted score corresponding to the print image to be predicted further includes: Obtain the value range of each waveform parameter; All values ​​within the given range are combined using Cartesian products to form a multidimensional grid; each dimension of the multidimensional grid corresponds to a waveform parameter, and each grid point corresponds to a set of waveform parameter combinations; each set of waveform parameter combinations generates a waveform to be filtered. The grid search method is used to traverse the grid points of the multidimensional grid, and the test image is printed according to the waveform to be screened corresponding to the grid point to obtain the print image to be predicted.

[0009] Optionally, the regression head structure is a linear layer, which outputs a single floating-point number.

[0010] Optionally, the backbone network is the first 17 layers of a ResNet18 convolutional neural network.

[0011] Compared with existing technologies, the present invention provides an automatic waveform selection method for printers, comprising: acquiring historical printed image data; determining the estimated score of each historical printed image using a crowdsourced Bradley-Terry model algorithm based on the historical printed image data; constructing an initial model for a score prediction model; the last layer of the initial model for a score prediction model is a regression head structure; training the initial model for a score prediction model based on the estimated score and historical printed images to obtain a score prediction model; feeding the printed image to be predicted into the score prediction model to predict the predicted score corresponding to the printed image to be predicted; the printed image to be predicted is obtained by printing a test image according to the waveform to be selected; sorting the waveforms to be selected according to the predicted scores, and determining the top M waveforms to be selected as target waveforms. This application uses a crowdsourced Bradley-Terry model algorithm to score historical printed images. It eliminates the need to train professional scorers and can quickly collect a dataset of printed images and their corresponding scores. This dataset is then used to train an initial model for score prediction. In practical applications, this score prediction model can be used to score printed images and sort the waveforms to be screened, selecting the highest quality waveforms. This eliminates the need for manual inspection and scoring of each printed image, reducing labor costs and improving waveform screening efficiency.

[0012] Secondly, the present invention provides an automatic waveform screening device for printers, comprising: The historical print image data acquisition module is used to acquire historical print image data; The scoring dataset determination module is used to determine the scoring estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm. The initial model building module for the score prediction model is used to build the initial model of the score prediction model; the last layer of the initial model of the score prediction model is a regression head structure. The model training module is used to train the initial model of the score prediction model based on the score estimate and the historical printed image to obtain the score prediction model; The score prediction module is used to input the print image to be predicted into the score prediction model and predict the score corresponding to the print image to be predicted; the print image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveform sorting module is used to sort the waveforms to be screened according to the predicted scores, and determine the top M waveforms to be screened as target waveforms, where M is a positive integer greater than 0.

[0013] Thirdly, the present invention provides an automatic waveform screening device for printers, comprising: Communication unit / communication interface, used to acquire historical printed image data; A processing unit / processor is used to determine the score estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm; Construct an initial model for the score prediction model; the last layer of the initial model for the score prediction model is a regression head structure. The initial model of the score prediction model is trained based on the score estimate and the historical printed image to obtain the score prediction model; The printed image to be predicted is fed into the score prediction model to predict the predicted score corresponding to the printed image; the printed image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveforms to be screened are sorted according to the predicted scores, and the top M waveforms to be screened are determined as target waveforms, where M is a positive integer greater than 0.

[0014] Fourthly, the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the automatic waveform selection method for printers.

[0015] Compared with the prior art, the beneficial effects of the second aspect device solution, the third aspect equipment solution, and the fourth aspect computer-readable storage medium solution provided by the present invention are the same as the beneficial effects of the automatic waveform screening method for printers described in the above technical solutions, and will not be repeated here. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 A flowchart of an automatic waveform selection method for printers provided by the present invention; Figure 2 This is a schematic diagram of the structure of an automatic waveform screening device for printers provided by the present invention; Figure 3 This is a schematic diagram of the structure of an automatic waveform screening device for printers provided by the present invention. Detailed Implementation

[0017] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" are not necessarily different.

[0018] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0019] In this invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, a combination of a and b, a combination of a and c, a combination of b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0020] Before introducing the embodiments of the present invention, the relevant terms involved in the embodiments of the present invention are first defined as follows: The Crowd-Bradley-Terry model (abbreviated as Crowd-BT model) is an extended application of the Bradley-Terry model in the crowdsourcing scenario. Its core is to objectively infer the relative merits or quantitative scores of multiple entities by aggregating pairwise comparison data from a large number of ordinary participants.

[0021] Cartesian product combination refers to the set of all possible ordered pairs formed by taking one element from each of multiple sets. It can systematically generate all possible combinations between multiple sets without omitting any combination.

[0022] When debugging a printer's waveform, it's necessary to collect numerous pairs of data—printed images and waveforms. Currently, this relies entirely on manual inspection of the images to determine their perfection, selecting the highest-quality waveforms for printing. Manually judging printed images requires a high level of expertise, necessitating extensive training of such personnel and time for the transmission of experience. The long waveform selection cycle and high labor costs ultimately lead to lengthy printer debugging times.

[0023] To address the aforementioned problems, this invention provides a method, apparatus, device, and medium for automatic waveform selection in printers, which will be described below with reference to the accompanying drawings.

[0024] See Figure 1 The present invention provides an automatic waveform selection method for printers, comprising the following steps: Step 100: Obtain historical printed image data; The historical printed image data includes: historical printed images, image numbers, and a comparison result set. The comparison result set includes the number of each group of historical printed images, the winning image number corresponding to each group of historical printed images, and the scorer's number. The collection process of the comparison result set is as follows: the collected printed images are uniformly named, totaling a images, maintaining a uniform size and resolution. Assuming there are m scorers, each person is randomly assigned at least a / m × 2 printed images, ensuring that each image is viewed by at least two people. Each scorer only needs to complete the following task: display two images in each round and select the one with higher quality. For example, as shown in Table 1, the scorer numbers include 001, 002, and 003. 001 selected image number 12 from the combination of image numbers 5 and 12; 002 selected image number 3 from the combination of image numbers 2 and 3; and 003 selected image number 9 from the combination of image numbers 13 and 9.

[0025] Table 1 Example of Comparison Result Set Step 200: Based on the historical printed image data, use the crowdsourcing Bradley-Terry model algorithm to determine the score estimate for each historical printed image; Specifically, step 200 includes the following steps: Step 210: Use formula (1): (1) Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, The probability of winning, i.e., the scorer's probability. The image is considered Superior to image The probability, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; Specifically, the score estimate for each historical print image can be obtained by inputting the scorer's number, the winner's image number, and the numbers of the two historical print images in each group into formula (1).

[0026] Step 220: With the goal of maximizing the likelihood function, the target parameters of the likelihood function are iteratively updated using an optimization algorithm until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the credibility of the rater, and the rater's preference bias.

[0027] You can use MATLAB to maximize the likelihood function.

[0028] Step 300: Construct the initial model of the score prediction model; the last layer of the initial model of the score prediction model is a regression head structure; Alternatively, the regression head structure is a linear layer that outputs a single floating-point number. The initial model of the score prediction model includes a backbone network, which is the first 17 layers of a ResNet18 convolutional neural network.

[0029] The ResNet18 convolutional neural network was chosen because of its deep residual connection structure, which effectively captures morphological details in ink spot images and improves training stability, making it suitable for small-sized image processing. Replacing the classification head in the last layer of the ResNet18 convolutional neural network with a regression head structure allows the model to output continuous quality scores, accurately reflecting the image quality level. This improvement combines robustness and efficiency, making it particularly suitable for automated image scoring and evaluation scenarios with high subjective factors. The regression head structure uses a linear layer to output a single floating-point number, better capturing subtle differences in image quality and maintaining consistency with human subjective scoring results. The regression head structure has stronger numerical expressive power, making it suitable for automated quality scoring.

[0030] Step 400: Train the initial model of the score prediction model based on the score estimate and the historical printed image to obtain the score prediction model; Specifically, step 400 can be implemented based on the following steps: Step 410: Input the historical printed image into the initial model of the score prediction model. The backbone network of the ResNet18 convolutional neural network extracts the morphological features in the historical printed image. The regression head structure processes the morphological features to obtain training prediction values.

[0031] Step 420: Calculate the mean squared error loss between the training prediction value and the score estimate, and adjust the parameters of the initial model of the score prediction model according to the mean squared error loss to obtain the score prediction model.

[0032] Mean squared error loss, also known as quadratic loss or L2 loss, is commonly used in regression prediction tasks. The mean squared error function measures the quality of a model by calculating the square of the distance between the predicted and actual values; that is, the closer the predicted and actual values ​​are, the smaller their mean squared error.

[0033] Specifically, formula (2) is used: (2) Calculate the mean squared error loss between the training prediction and the score estimate; in, For mean square error loss, The number of ratings estimated, For rating and valuation, These are the training prediction values.

[0034] Step 500: Input the print image to be predicted into the score prediction model to predict the predicted score corresponding to the print image; the print image to be predicted is obtained by printing the test image according to the waveform to be screened; Step 600: Sort the waveforms to be screened according to the predicted scores, and determine the top M waveforms to be screened as target waveforms, where M is a positive integer greater than 0.

[0035] As an optional approach, the step of feeding the printed image to be predicted into the score prediction model to predict the predicted score corresponding to the printed image to be predicted further includes: Obtain the value range of each waveform parameter; All values ​​within the given range are combined using Cartesian products to form a multidimensional grid; each dimension of the multidimensional grid corresponds to a waveform parameter, and each grid point corresponds to a set of waveform parameter combinations; each set of waveform parameter combinations generates a waveform to be filtered. The grid search method is used to traverse the grid points of the multidimensional grid, and the test image is printed according to the waveform to be screened corresponding to the grid point to obtain the print image to be predicted.

[0036] The above method can be used to quickly obtain a batch of waveforms and corresponding printed images.

[0037] Figure 1 The method described above uses the crowdsourced Bradley-Terry model algorithm to score historical printed images. It does not require training professional scorers and can quickly collect a dataset of printed images and their corresponding scores. This dataset is used to train the initial model of the score prediction model. In practical applications, the score prediction model can be used to score printed images and sort the waveforms to be screened, selecting the waveforms with the highest quality. This eliminates the need for manual inspection and scoring of each printed image, reducing labor costs and improving waveform screening efficiency.

[0038] The embodiments of the present invention can divide functional modules according to the above method examples. For example, each function can be divided into its own functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in the embodiments of the present invention is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0039] When dividing each function into modules according to its corresponding function. Figure 2 A schematic diagram of the structure of an automatic waveform screening device for printers provided by the present invention is shown. Figure 2 As shown, the device includes: Historical print image data acquisition module 201 is used to acquire historical print image data; The scoring dataset determination module 202 is used to determine the scoring estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm. The initial model construction module 203 for the score prediction model is used to construct the initial model of the score prediction model; the last layer of the initial model of the score prediction model is a regression head structure. Model training module 204 is used to train the initial model of the score prediction model based on the score estimate and the historical printed image to obtain the score prediction model; The score prediction module 205 is used to input the print image to be predicted into the score prediction model and predict the predicted score corresponding to the print image to be predicted; the print image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveform sorting module 206 is used to sort the waveforms to be screened according to the predicted scores, and determine the top M waveforms to be screened as target waveforms, where M is a positive integer greater than 0.

[0040] Optionally, the historical printed image data includes: historical printed images, image numbers, and a comparison result set; the comparison result set includes the number of each group of historical printed images, the winner image number corresponding to each group of historical printed images, and the scorer number; the scoring dataset determination module 202 may specifically include: Likelihood function building blocks are used to employ the formula: Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, For the probability of winning, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; The optimization unit is used to iteratively update the target parameters of the likelihood function with the goal of maximizing the likelihood function, until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the rater's credibility, and the rater's preference bias.

[0041] Optionally, the initial model of the score prediction model further includes a backbone network, and the model training module 204 may specifically include: The training prediction value determination unit is used to input the historical printed image into the initial model of the score prediction model; the backbone network extracts morphological features from the historical printed image; the regression head structure processes the morphological features to obtain the training prediction value. The mean squared error loss calculation unit is used to calculate the mean squared error loss between the training predicted value and the score estimate, and adjust the parameters of the initial model of the score prediction model according to the mean squared error loss to obtain the score prediction model.

[0042] Optionally, the mean square error loss calculation unit can be specifically used for: Formula used: Calculate the mean squared error loss between the training prediction and the score estimate; in, For mean square error loss, The number of ratings estimated, For rating and valuation, These are the training prediction values.

[0043] Optionally, the device may further include an image acquisition module to be predicted, which may specifically include: The value range acquisition unit is used to obtain the value range of each waveform parameter; A multidimensional grid generation unit is used to perform Cartesian product combinations on all values ​​within the specified range to form a multidimensional grid; each dimension of the multidimensional grid corresponds to a waveform parameter, and each grid point corresponds to a set of waveform parameter combinations; each set of waveform parameter combinations generates a waveform to be filtered. The printing unit is used to traverse the grid points of the multidimensional grid using a grid search method, and print the test image according to the waveform to be screened corresponding to the grid point to obtain the print image to be predicted.

[0044] Optionally, the regression head structure is a linear layer, which outputs a single floating-point number.

[0045] Optionally, the backbone network is the first 17 layers of a ResNet18 convolutional neural network.

[0046] The above mainly describes the solutions provided by the embodiments of the present invention from the perspective of the interaction between various modules. It is understood that, in order to achieve the above functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, the present invention can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0047] When using the corresponding integrated unit Figure 3 This diagram illustrates the structure of an automatic waveform screening device for printers provided by the present invention. Figure 3 As shown, the device includes: a communication unit / communication interface for acquiring historical printed image data; A processing unit / processor is used to determine the score estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm; Construct an initial model for the score prediction model; the last layer of the initial model for the score prediction model is a regression head structure. The initial model of the score prediction model is trained based on the score estimate and the historical printed image to obtain the score prediction model; The printed image to be predicted is fed into the score prediction model to predict the predicted score corresponding to the printed image; the printed image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveforms to be screened are sorted according to the predicted scores, and the top M waveforms to be screened are determined as target waveforms, where M is a positive integer greater than 0.

[0048] The processing unit can be a processor or controller, such as a Central Processing Unit (CPU), a general-purpose processor, a Digital Signal Processor (DSP), an Application-Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The communication module can be a transceiver, transceiver circuitry, or communication interface, etc. The storage module can be a memory.

[0049] like Figure 3 As shown, the processor described above can be a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. The communication interface described above can be one or more. The communication interface can use any transceiver-like device for communicating with other devices or communication networks.

[0050] like Figure 3 As shown, the device may also include a communication line. The communication line may include a path for transmitting information between the components.

[0051] Optional, such as Figure 3 As shown, the device may further include a memory. The memory stores computer execution instructions for carrying out the present invention, and the execution is controlled by a processor. The processor executes the computer execution instructions stored in the memory, thereby implementing the method provided in the embodiments of the present invention.

[0052] like Figure 3As shown, the memory can be read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions, random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed discs, laser discs, optical discs, digital universal discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited to these. The memory can exist independently and be connected to the processor via communication lines. The memory can also be integrated with the processor.

[0053] Optionally, the computer execution instructions in the embodiments of the present invention may also be referred to as application code, and the embodiments of the present invention do not specifically limit this.

[0054] In a specific implementation, as one example, such as Figure 3 As shown, a processor may include one or more CPUs, such as Figure 3 CPU0 and CPU1 in the CPU.

[0055] In a specific implementation, as one example, such as Figure 3 As shown, the terminal device may include multiple processors, such as Figure 3 The processors in the system. Each of these processors can be a single-core processor or a multi-core processor.

[0056] On the one hand, a computer-readable storage medium is provided, which stores instructions that, when executed, implement an automatic waveform selection method for printers.

[0057] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present invention are performed entirely or partially. The computer can be a general-purpose computer, a special-purpose computer, a computer network, a terminal, a user equipment, or other programmable device. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; it can also be an optical medium, such as a digital video disc (DVD); or it can be a semiconductor medium, such as a solid-state drive (SSD).

[0058] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims in carrying out the claimed invention. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0059] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely exemplary descriptions of the invention as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method for automatically selecting waveforms in a printer, characterized in that, include: Retrieve historical printed image data; Based on the historical printed image data, the crowdsourced Bradley-Terry model algorithm is used to determine the score estimate for each historical printed image; The historical printed image data includes: historical printed images, image numbers, and a comparison result set; the comparison result set includes the number of each group of historical printed images, the winner image number corresponding to each group of historical printed images, and the scorer number; the step of determining the score estimate for each historical printed image using the crowdsourcing Bradley-Terry model algorithm based on the historical printed image data includes: Formula used: ； Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, For the probability of winning, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; With the goal of maximizing the likelihood function, an optimization algorithm is used to iteratively update the target parameters of the likelihood function until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the credibility of the rater, and the rater's preference bias. Construct an initial model for the score prediction model; the last layer of the initial model for the score prediction model is a regression head structure. The initial model of the score prediction model is trained based on the score estimate and the historical printed image to obtain the score prediction model; The printed image to be predicted is fed into the score prediction model to predict the predicted score corresponding to the printed image; the printed image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveforms to be screened are sorted according to the predicted scores, and the top M waveforms to be screened are determined as target waveforms, where M is a positive integer greater than 0.

2. The automatic waveform selection method for printers according to claim 1, characterized in that, The initial model of the score prediction model also includes a backbone network. Training the initial model based on the score estimate and the historical printed image to obtain the score prediction model includes: The historical printed image is input into the initial model of the score prediction model, and the backbone network extracts morphological features from the historical printed image; the regression head structure processes the morphological features to obtain training prediction values. Calculate the mean squared error loss between the training prediction value and the score estimate, and adjust the parameters of the initial model of the score prediction model according to the mean squared error loss to obtain the score prediction model.

3. The automatic waveform selection method for printers according to claim 2, characterized in that, The calculation of the mean squared error loss between the training prediction value and the score estimate includes: Formula used: ； Calculate the mean squared error loss between the training prediction and the score estimate; in, For mean square error loss, The number of ratings estimated, For rating and valuation, These are the training prediction values.

4. The automatic waveform selection method for printers according to claim 1, characterized in that, The step of feeding the print image to be predicted into the score prediction model to predict the predicted score corresponding to the print image to be predicted also includes: Obtain the value range of each waveform parameter; All values ​​within the given range are combined by Cartesian product to form a multidimensional grid; each dimension of the multidimensional grid corresponds to a waveform parameter, and each grid point corresponds to a set of waveform parameter combinations; each set of waveform parameter combinations generates a waveform to be filtered. The grid search method is used to traverse the grid points of the multidimensional grid, and the test image is printed according to the waveform to be screened corresponding to the grid point to obtain the print image to be predicted.

5. The automatic waveform selection method for printers according to claim 1, characterized in that, The regression head structure is a linear layer, and the linear layer outputs a single floating-point number.

6. The automatic waveform selection method for printers according to claim 2, characterized in that, The backbone network is the first 17 layers of a ResNet18 convolutional neural network.

7. An automatic waveform screening device for printers, characterized in that, include: The historical print image data acquisition module is used to acquire historical print image data; The scoring dataset determination module is used to determine the scoring estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm. The historical printed image data includes: historical printed images, image numbers, and a comparison result set; the comparison result set includes the number of each group of historical printed images, the winner image number corresponding to each group of historical printed images, and the scorer number; the step of determining the score estimate for each historical printed image using the crowdsourcing Bradley-Terry model algorithm based on the historical printed image data includes: Formula used: ； Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, For the probability of winning, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; With the goal of maximizing the likelihood function, an optimization algorithm is used to iteratively update the target parameters of the likelihood function until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the credibility of the rater, and the rater's preference bias. The initial model building module for the score prediction model is used to build the initial model of the score prediction model; the last layer of the initial model of the score prediction model is a regression head structure. The model training module is used to train the initial model of the score prediction model based on the score estimate and the historical printed image to obtain the score prediction model; The score prediction module is used to input the print image to be predicted into the score prediction model and predict the score corresponding to the print image to be predicted; the print image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveform sorting module is used to sort the waveforms to be screened according to the predicted scores, and determine the top M waveforms to be screened as target waveforms, where M is a positive integer greater than 0.

8. An automatic waveform screening device for printers, characterized in that, include: Communication unit / communication interface, used to acquire historical printed image data; A processing unit / processor is used to determine the score estimate of each historical printed image based on the historical printed image data using a crowdsourced Bradley-Terry model algorithm; The historical printed image data includes: historical printed images, image numbers, and a comparison result set; the comparison result set includes the number of each group of historical printed images, the winner image number corresponding to each group of historical printed images, and the scorer number; the step of determining the score estimate for each historical printed image using the crowdsourcing Bradley-Terry model algorithm based on the historical printed image data includes: Formula used: ； Obtain the likelihood function; in, , , For rating and valuation, For the scorer's number, To ensure the credibility of the raters, For the rater's preference bias, For the probability of winning, Assign numbers to the two historical printed images in each group of historical printed images. For the first The first scorer's The score of each historical printed image; For rating and valuation, Number the winner's image; With the goal of maximizing the likelihood function, an optimization algorithm is used to iteratively update the target parameters of the likelihood function until the parameter change is less than the convergence threshold, thus obtaining the optimal solution of the target parameters; the target parameters include the rating estimate, the credibility of the rater, and the rater's preference bias. Construct an initial model for the score prediction model; the last layer of the initial model for the score prediction model is a regression head structure. The initial model of the score prediction model is trained based on the score estimate and the historical printed image to obtain the score prediction model; The printed image to be predicted is fed into the score prediction model to predict the predicted score corresponding to the printed image; the printed image to be predicted is obtained by printing the test image according to the waveform to be screened; The waveforms to be screened are sorted according to the predicted scores, and the top M waveforms to be screened are determined as target waveforms, where M is a positive integer greater than 0.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a processor, implement the automatic waveform selection method for printers as described in any one of claims 1-6.

Images