A method for automatically generating printed graphics and text on a sports skateboard

Abstract

The present invention relates to a method and system for automatically generating printed images on a sports skateboard, and relates to the technical field of skateboard surface printing. The system comprises: an image management module for establishing an image database for storing multiple template images; a graphic generation module for generating an image to be transferred based on user graphic requirements and the image database; a parameter determination module for determining thermal transfer parameters based on relevant information of the sports skateboard and image features of the image to be transferred, the parameters including at least thermal transfer temperature, thermal transfer pressure, thermal transfer time and thermal transfer speed; a graphic transfer module for controlling a thermal transfer device to print the image to be transferred on the skateboard surface based on the thermal transfer parameters of the image to be transferred; and a quality inspection module for performing quality assessment based on the image of the skateboard surface after printing and the image features of the image to be transferred. The system has the advantages of realizing personalized and automated design and printing of printed images on sports skateboards.

Description

Technical Field

[0001] The present invention relates to the technical field of skateboard surface printing, and in particular to a method and system for automatically generating printed graphics and text on a sports skateboard. Background Art

[0002] Skateboards are becoming more and more popular in people's lives today, especially among young people. The pursuit of skateboards is not only limited to their functions and quality, but the design and pursuit of appearance style are also very important. Therefore, when users use skateboards, they will print various cartoons or patterns on the inside of the skateboards to highlight their personal personality and enhance their appreciation value.

[0003] The thermal transfer process requires precise control of parameters such as temperature, pressure, time, and speed. The selection and adjustment of these parameters directly impacts the quality of the printed product. Misaligned parameters can easily lead to issues such as discoloration, scratches, and fading. The complexity of parameter control requires operators to possess high levels of professional skills and experience, increasing both the technical complexity and costs of the production process.

[0004] Therefore, it is necessary to provide a method and system for automatically generating printed graphics on a sports skateboard, so as to realize the design and printing of personalized and automated printed graphics on a sports skateboard. Summary of the Invention

[0005] The present invention provides an automatic generation system for printed images and texts on a sports skateboard, comprising: an image management module for establishing an image database, wherein the image database is used to store multiple template images; an image and text generation module for obtaining user image and text requirements, and generating an image to be transferred based on the user image and text requirements and the image database; a parameter determination module for obtaining relevant information of the sports skateboard, extracting image features of the image to be transferred, and determining thermal transfer parameters based on the relevant information of the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters include at least thermal transfer temperature, thermal transfer pressure, thermal transfer time and thermal transfer speed; an image and text transfer module for controlling a thermal transfer device to print the image to be transferred on the skateboard surface based on the thermal transfer parameters of the image to be transferred; a quality detection module for obtaining an image of the skateboard surface after printing, and performing quality assessment based on the image of the skateboard surface after printing and the image features of the image to be transferred.

[0006] Furthermore, the image management module establishes an image database, including: for each of the template images, extracting the image label of the template image; for any two of the template images, calculating the label similarity of the image labels of the two template images; based on any two of the template images, clustering the multiple template images to determine multiple image clusters; and establishing the image database based on the multiple image clusters.

[0007] Furthermore, the image management module extracts the image label of the template image, including: extracting the image label of the template image through a label generation model, wherein the image label of the template image includes at least object labels, scene labels, event labels, emotion labels and color feature labels.

[0008] Furthermore, the image and text generation module generates an image to be transferred based on the user image and text requirements and the image database, including: performing semantic understanding of the user image and text requirements to determine image requirement information and text requirement information; determining a target template image from the image database based on the image requirement information; generating an intermediate image based on the image requirement information and the target template image through an image generation model; and generating an image to be transferred based on the text requirement information and the intermediate image through an image and text fusion model.

[0009] Furthermore, the image-text generation module determines the target template image from the image database based on the image requirement information, including: determining the cluster matching value of each image cluster based on the image requirement keywords and the image label of the template image included in each image cluster; determining the target image cluster from the multiple image clusters based on the cluster matching degree of each image cluster; determining the image matching value of each template image included in the target image cluster based on the image requirement keywords and the image label of each template image included in the target image cluster; and determining the target template image from the target image cluster based on the image matching value of each template image included in the target image cluster.

[0010] Furthermore, the parameter determination module extracts image features of the image to be transferred, including: obtaining adjacent pixel matrix feature parameters of each pixel of the image to be transferred, and determining edge pixels based on the adjacent pixel matrix feature parameters of each pixel; for non-edge pixels of the image to be transferred, clustering the non-edge pixels of the image to be transferred based on the RGB value of each non-edge pixel to determine multiple non-edge pixel clusters; determining color distribution features of the image to be transferred based on the multiple non-edge pixel clusters; determining detail level features of the image to be transferred based on the edge pixels; converting the image to be transferred into a grayscale image to be transferred; and extracting the grayscale co-occurrence matrix and autocorrelation function of the grayscale image to be transferred.

[0011] Furthermore, the parameter determination module is also used to: establish a processing sample database, wherein the processing sample database is used to store multiple processing samples, and the processing samples include image features of the sample transfer image and sample thermal transfer parameters; the parameter determination module determines the thermal transfer parameters based on the relevant information of the motion skateboard and the image features of the image to be transferred, including: determining the target processing sample from the processing sample database based on the image features of the image to be transferred; and determining the thermal transfer parameters based on the target processing sample and the image features of the image to be transferred through a parameter generation model.

[0012] Furthermore, the quality detection module is also used to: obtain real-time printing parameters in the process of controlling the thermal transfer device to print the image to be transferred on the surface of the skateboard; and determine the fault information of the thermal transfer device based on the real-time printing parameters of multiple consecutive printing time points through a fault analysis model.

[0013] Furthermore, the quality detection module performs quality assessment based on the image of the printed skateboard surface and the image features of the image to be transferred, including: extracting color distribution features, detail level features, grayscale co-occurrence matrix and autocorrelation function of the image of the printed skateboard surface; calculating the image similarity between the image of the printed skateboard surface and the image to be transferred based on the color distribution features, detail level features, grayscale co-occurrence matrix and autocorrelation function of the image of the printed skateboard surface and the color distribution features, detail level features, grayscale co-occurrence matrix and autocorrelation function of the image of the printed skateboard surface; and performing quality assessment based on the image similarity.

[0014] The present invention provides a method for automatically generating printed images and texts on a sports skateboard, comprising: establishing an image database, wherein the image database is used to store multiple template images; obtaining user image and text requirements, and generating an image to be transferred based on the user image and text requirements and the image database; obtaining relevant information of the sports skateboard, and extracting image features of the image to be transferred; determining thermal transfer parameters based on the relevant information of the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters include at least thermal transfer temperature, thermal transfer pressure, thermal transfer time, and thermal transfer speed; controlling a thermal transfer device to print the image to be transferred on the skateboard surface based on the thermal transfer parameters of the image to be transferred; obtaining an image of the skateboard surface after printing, and performing a quality assessment based on the image of the skateboard surface after printing and the image features of the image to be transferred.

[0015] Compared with the prior art, the method for automatically generating printed graphics and text on a sports skateboard provided in this specification has at least the following beneficial effects:

[0016] 1. Allowing users to generate transfer images based on their desired graphics and text, each skateboard can be unique, satisfying consumers' desire for personalized products. The entire process, from image generation to parameter determination and transfer, is highly automated, reducing manual intervention and improving production efficiency. In particular, the parameter determination module intelligently analyzes the skateboard's information and the characteristics of the image to be transferred, automatically adjusting the thermal transfer parameters to ensure accuracy and efficiency. The quality inspection module compares the printed skateboard surface image with the characteristics of the image to be transferred to assess quality, ensuring high clarity and accuracy of the transferred image and improving overall product quality. This automated and intelligent production process reduces human error and waste, lowering production costs. Furthermore, the precise control of thermal transfer parameters avoids material waste and defective products caused by improper parameters.

[0017] 2. The image management module builds an organized and efficient image database by extracting image tags from template images, calculating tag similarity, and clustering them into image clusters. This not only facilitates image retrieval and management but also improves the matching accuracy and efficiency of the image-text generation module. The image-text generation module semantically understands the user's image-text requirements and, in combination with an image generation model and an image-text fusion model, generates images for transfer that both meet the user's intent and offer good visual quality. This image-text fusion approach makes the generated images more vivid and interesting, attracting more consumers. The parameter determination module extracts the image features of the image to be transferred and, combined with relevant information about the sports skateboard, accurately determines the thermal transfer parameters. This ensures the efficiency and accuracy of the transfer process and avoids transfer quality issues caused by inappropriate parameters. The quality inspection module compares the image features of the printed skateboard surface with the image to be transferred to perform quality assessment, ensuring the high clarity and accuracy of the transferred image. This helps improve overall product quality and enhance consumer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0019] Figure 1 This is a module diagram of a system for automatically generating printed graphics and text on a sports skateboard, as shown in one embodiment of the present application;

[0020] Figure 2 is a flow chart of extracting image features of an image to be transferred shown in one embodiment of the present application;

[0021] Figure 3This is a flow chart of a method for automatically generating printed graphics and text on a sports skateboard shown in one embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly introduces the drawings required for describing the embodiments.

[0023] Figure 1 This is a module diagram of a system for automatically generating printed graphics and text on a sports skateboard shown in one embodiment of the present application. Figure 1 As shown, a system for automatically generating printed images and texts on a sports skateboard surface may include an image management module, an image and text generation module, a parameter determination module, an image and text transfer module, and a quality detection module.

[0024] The image management module can be used to establish an image database, wherein the image database is used to store multiple template images.

[0025] In some embodiments, the image management module establishes an image database, including:

[0026] For each template image, extract the image label of the template image;

[0027] For any two template images, calculate the label similarity of the image labels of the two template images;

[0028] Based on any two template images, multiple template images are clustered to determine multiple image clusters;

[0029] An image database is established based on multiple image clusters.

[0030] In some embodiments, the image management module extracts the image tag of the template image, including:

[0031] The image label of the template image is extracted through a label generation model, wherein the image label of the template image includes at least an object label, a scene label, an event label, an emotion label and a color feature label, wherein the label generation model can be a convolutional neural network (CNN) model.

[0032] Specifically, object classification involves classifying an image based on whether it contains a specific object or objects. This approach focuses on the presence of a predefined object category within an image, without necessarily requiring precise segmentation of the object. For example, a label generation model can determine whether an image contains a cat, dog, car, or other object.

[0033] Scene classification classifies images based on the physical environment or setting in which they were captured. This classification method focuses on understanding the overall environment or context of the image, such as a beach, a forest, or a city street. By identifying the scene within an image, the label generation model can provide a broader classification and understanding of the image content.

[0034] Event classification focuses on whether a specific event or activity has occurred in an image. This classification method requires identifying actions, interactions between people, and possible contextual information in the image to determine whether the image depicts an event such as a wedding, a sports event, or a traffic accident.

[0035] Sentiment classification is a more complex image classification method that attempts to classify images based on the emotions or moods conveyed by their content. This classification method not only focuses on the objects or scenes in the image, but also considers how factors such as color, composition, and lighting combine to create a specific emotional atmosphere. For example, an image might be classified as "warm," "sad," or "excited."

[0036] The label similarity of the image labels of two template images can be calculated based on the following formula:

[0037]

[0038] Among them, S (i,j) is the label similarity between the image labels of the i-th template image and the j-th template image, ε ((i,j),n) is the consistency parameter of the nth image label of the i-th template image and the j-th template image. When the nth image label of the i-th template image is the same as the nth image label of the j-th template image, ε ((i,j),n) is 1, when the nth image label of the i-th template image is different from the nth image label of the j-th template image, ε ((i,j),n) is 0, and N is the total number of image labels.

[0039] Multiple template images can be clustered based on the following process to determine multiple image clusters:

[0040] S11. For each template image, based on the label similarity between the template image and each other template image, calculate a first cluster matching parameter of the template image;

[0041] S12, taking the template image whose first cluster matching parameter is greater than the first cluster matching parameter threshold as the cluster center;

[0042] S13, for each template image that is not a cluster center, assign the template image to the cluster center with the largest label similarity;

[0043] S14. For each image cluster, calculate the number of template images included in the image cluster, and cancel the cluster centers of image clusters whose number of template images is less than a threshold value;

[0044] S15. For each image cluster including a number of template images greater than or equal to a threshold number, calculate a second cluster matching parameter for the template image based on label similarity between the template image and each other template image included in the image cluster, and use the template image with the largest second cluster matching parameter as a new cluster center.

[0045] S16: Determine whether the cluster center has changed. If so, execute S13; if not, complete clustering.

[0046] For example, the first cluster matching parameter of the template image can be calculated based on the following formula:

[0047]

[0048] Among them, M (i,1) is the first cluster matching parameter of the i-th template image, S (i,n) is the label similarity between the image labels of the i-th template image and the n-th template image, δ1 is a preset parameter, and δ1 is greater than 0.

[0049] The image and text generation module can be used to obtain user image and text requirements, and generate images to be transferred based on the user image and text requirements and the image database.

[0050] Specifically include:

[0051] Conduct semantic understanding of user image and text requirements to determine image and text requirement information;

[0052] Based on the image requirement information, a target template image is determined from the image database;

[0053] Generate an intermediate image based on the image requirement information and the target template image through an image generation model, wherein the image generation model may be a Generative Adversarial Network (GAN) model;

[0054] The image to be transferred is generated based on the text requirement information and the intermediate image through the image-text fusion model, wherein the image-text fusion model can be a Generative Adversarial Networks (GAN) model.

[0055] For example, users express their graphic and text needs through text descriptions (for example, "I want a summer landscape with an ocean, a beach, and a sunset") or by uploading reference images. They may also include specific requirements such as color, style (for example, oil painting, watercolor, sketching, etc.), emotional atmosphere (for example, warmth, tranquility, etc.), and characteristics of the text to be displayed in the transferred image (for example, specific characters, fonts, character size, etc.).

[0056] Use natural language processing (NLP) technology to parse the text entered by the user and identify keywords (for example, "ocean", "beach", "sunset", etc.) and modifiers (for example, "summer", "warm", etc.).

[0057] In some embodiments, the image-text generation module determines a target template image from an image database based on the image requirement information, including:

[0058] determining a cluster matching value for each image cluster based on the image requirement keywords and the image labels of the template images included in each image cluster;

[0059] Based on the cluster matching degree of each image cluster, a target image cluster is determined from the multiple image clusters, for example, an image cluster whose cluster matching value is greater than a cluster matching value threshold is taken as the target image cluster;

[0060] determining an image matching value of each template image included in the target image cluster based on the image requirement keyword and the image label of each template image included in the target image cluster;

[0061] Based on the image matching value of each template image included in the target image cluster, a target template image is determined from the target image cluster. For example, a template image in the target image cluster whose image matching value is greater than an image matching value threshold is used as the target template image.

[0062] Specifically, for each image cluster, multiple template images are sampled from the image cluster, image matching values ​​of the template images are calculated, and based on the image matching values ​​of the sampled multiple template images, a cluster matching value of the image cluster is calculated.

[0063] For example, the cluster matching value of an image cluster can be calculated based on the following formula:

[0064]

[0065]

[0066] Among them, M (k,2) is the cluster matching value of the kth image cluster, M (k,m) is the image matching value of the mth template image sampled from the kth image cluster, M is the total number of template images sampled from the kth image cluster, ε ((k,m),n)is the consistency parameter of the image requirement keyword and the mth template image sampled from the kth image cluster at the nth image label. When the image requirement keyword and the mth template image sampled from the kth image cluster match at the nth image label, ε ((k,m),n) is 1, when the image requirement keyword and the mth template image sampled from the kth image cluster do not match the nth image label, ε ((k,m),n) is 0.

[0067] The parameter determination module can be used to obtain relevant information of the sports skateboard (for example, material, size, characteristics of the thermal transfer film, etc.), extract image features of the image to be transferred, and determine the thermal transfer parameters based on the relevant information of the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters include at least thermal transfer temperature, thermal transfer pressure, thermal transfer time and thermal transfer speed.

[0068] Specifically, the temperature for thermal transfer is generally set between 140°C and 190°C, but the specific temperature needs to be adjusted based on the material of the slide, the type of thermal transfer film, and the image characteristics of the image to be transferred. The pressure for thermal transfer is generally set between 4 and 6 kg / cm2. However, this also needs to be adjusted based on the material of the slide, the characteristics of the thermal transfer film, and the image characteristics of the image to be transferred. If the pressure is too low, the thermal transfer film will not be able to fully contact the surface of the slide, resulting in a weak transfer; if the pressure is too high, the slide surface may be compressed and deformed, affecting the accuracy and aesthetics of the pattern. The thermal transfer time is generally determined by factors such as temperature, pressure, and the material of the slide, and is usually between a few seconds and tens of seconds. Too short a time may result in incomplete transfer, while too long a time may cause the slide surface to overheat, affecting the quality of the pattern. The marking or printing speed of the thermal transfer equipment can fluctuate within a certain range. For example, when printing labels or simple patterns, some thermal transfer equipment can produce hundreds of labels or patterns per minute. However, when performing thermal transfer of large areas and complex patterns, the speed may be relatively slow. The thermal transfer speed is determined based on the relevant information of the sports skateboard and the image features of the image to be transferred to achieve the best thermal transfer effect.

[0069] Figure 2 This is a flow chart of extracting image features of an image to be transferred, as shown in one embodiment of the present application. Figure 2 As shown, in some embodiments, the parameter determination module extracts image features of the image to be transferred, including:

[0070] Obtaining adjacent pixel matrix characteristic parameters of each pixel of the image to be transferred, and determining edge pixels based on the adjacent pixel matrix characteristic parameters of each pixel;

[0071] For non-edge pixels of the image to be transferred, clustering the non-edge pixels of the image to be transferred based on the RGB value of each non-edge pixel to determine a plurality of non-edge pixel clusters;

[0072] determining a color distribution feature of the image to be transferred based on a plurality of non-edge pixel clusters;

[0073] Determine the detail level features of the image to be transferred based on edge pixels;

[0074] converting the image to be transferred into a grayscale image to be transferred;

[0075] Extract the gray level co-occurrence matrix and autocorrelation function of the gray level image to be transferred.

[0076] Specifically, for each pixel of the image to be transferred, pixels whose pixel distance is less than the preset pixel distance can be used as the adjacent pixels of the pixel according to the preset pixel distance. The adjacent pixel matrix characteristic parameters of the pixel are calculated based on the following formula:

[0077]

[0078] Among them, σ e is the adjacent pixel matrix characteristic parameter of the e-th pixel, RGB (e,f) is the RGB value of the fth neighboring pixel of the eth pixel, and F is the total number of neighboring pixels of the eth pixel.

[0079] Pixels whose adjacent pixel matrix feature parameters are greater than the adjacent pixel matrix feature parameter threshold are regarded as edge pixels.

[0080] The non-edge pixels of the image to be transferred may be clustered based on the RGB value of each non-edge pixel using a K-means clustering algorithm to determine a plurality of non-edge pixel clusters.

[0081] The color distribution characteristics of the image to be transferred may include the number of non-edge pixel clusters, the RGB value mean of each non-edge pixel cluster, and the RGB value variance of the non-edge pixel cluster calculated based on the RGB value mean of each non-edge pixel cluster.

[0082] For edge pixels of the image to be transferred, the adjacent pixel matrix feature parameters of each edge pixel are clustered to determine multiple edge pixel clusters. The detail level features of the image to be transferred may include the number of edge pixel clusters, the ratio of edge pixels, etc.

[0083] The gray-level co-occurrence matrix is ​​a two-dimensional matrix whose size is determined by the number of grayscale levels after quantization. Each element P(i, j) in the matrix represents the frequency of the co-occurrence of a pixel with grayscale value i and a pixel with grayscale value j at a distance d and a direction θ. Four directions θ are usually considered: 0 degrees (horizontal), 90 degrees (vertical), 45 degrees, and 135 degrees. The distance d is a positive integer that represents the relative distance between pixel pairs. Traverse each pixel in the image, and for each pixel, find the corresponding pixel pair based on its grayscale value, the specified distance d, and the direction θ, and update the value of the corresponding position in the gray-level co-occurrence matrix. To obtain a probability distribution, it is usually necessary to divide each element in the gray-level co-occurrence matrix by the sum of all elements in the matrix to achieve normalization. A series of texture features can be extracted from the normalized gray-level co-occurrence matrix, including but not limited to: Energy: reflecting the uniformity of image texture; Contrast: reflecting the clarity of image texture and the depth of visual effect; Entropy: reflecting the complexity or information content of image texture; Correlation: reflecting the linear correlation between pixels in image texture; Inverse Difference Moment (IDM): reflecting the local homogeneity of image texture.

[0084] For two-dimensional images, the autocorrelation function can be defined as the similarity between the image and itself under different displacements. Specifically, let the original image be f(x,y), then the autocorrelation function R(u,v) can be expressed as a similarity measure between the images f(x,y) and f(x+u,y+v), where (u,v) is the displacement vector. Calculation process: Select a displacement range (the value range of u and v), which determines the resolution and computational complexity of the autocorrelation function. For each displacement (u,v), calculate the similarity between the original image f(x,y) and the displaced image f(x+u,y+v). This is usually achieved by calculating the sum of products (or normalized sum of products) of the two image regions. The similarity values ​​corresponding to all displacements are stored to form the output matrix of the autocorrelation function.

[0085] In some embodiments, the parameter determination module is further used to: establish a processing sample database, wherein the processing sample database is used to store a plurality of processing samples, and the processing samples include image features of the sample transfer image and sample thermal transfer parameters.

[0086] In some embodiments, the parameter determination module determines the thermal transfer parameters based on the relevant information of the motion skateboard and the image features of the image to be transferred, including:

[0087] determining a target processing sample from a processing sample database based on image features of the image to be transferred;

[0088] The thermal transfer parameters are determined based on the image features of the target processing sample and the image to be transferred through a parameter generation model, wherein the parameter generation model can be a convolutional neural network (CNN) model.

[0089] Specifically, for each processed sample, the image feature similarity between the image features of the sample transferred image of the processed sample and the image features of the image to be transferred can be calculated, and the processed sample with image feature similarity greater than the image feature similarity threshold is used as the target processed sample.

[0090] For example, image feature similarity can be calculated based on the following formula:

[0091]

[0092] Among them, S r is the image feature similarity between the image features of the sample transfer image of the rth processed sample and the image features of the image to be transferred, α1, α2, α3, α4, α5, α6 and α7 are weights, α1, α2, α3, α4, α5, α6 and α7 are all greater than 0, α1+α2+α3+α4+α5+α6+α7=1, δ2, δ3, δ4, δ5 and δ6 are preset parameters, and δ2, δ3, δ4, δ5 and δ6 are greater than 0, N (r,1) is the number of non-edge pixel clusters of the sample transfer image of the rth processed sample, N (c,1) is the number of non-edge pixel clusters of the image features to be transferred, S (r,1) is the cosine similarity between the RGB value mean sequence of the non-edge pixel cluster of the sample transfer image of the rth processed sample and the RGB value mean sequence of the non-edge pixel cluster of the image feature to be transferred. The RGB value mean sequence of the non-edge pixel cluster of the image is composed of the RGB value mean of each non-edge pixel cluster arranged from large to small according to the number of pixels included in the non-edge pixel cluster. σ (r,1) is the RGB value variance of the non-edge pixel cluster of the sample transfer image of the rth processed sample, σ (c,1) is the variance of the RGB values ​​of the non-edge pixel clusters of the image to be transferred, N (r,2) is the number of edge pixel clusters of the sample transfer image of the rth processed sample, N (c,2) is the number of edge pixel clusters of the image to be transferred, P (r,1) is the edge pixel ratio of the sample transfer image of the rth processed sample, P (c,1) is the edge pixel ratio of the edge pixel cluster of the image to be transferred, S (r,1) is the cosine similarity of the gray level co-occurrence matrix between the sample transfer image of the rth processed sample and the image to be transferred, S (r,2)is the similarity between the autocorrelation function of the sample transferred image of the rth processed sample and the image to be transferred.

[0093] The image transfer module can be used to control the thermal transfer device to print the image to be transferred on the surface of the skateboard based on the thermal transfer parameters of the image to be transferred.

[0094] The quality detection module can be used to obtain an image of the surface of the printed skateboard and perform quality assessment based on the image of the surface of the printed skateboard and image features of the image to be transferred.

[0095] In some embodiments, the quality detection module is further configured to:

[0096] Acquiring real-time printing parameters (e.g., real-time thermal transfer temperature, real-time thermal transfer pressure, real-time thermal transfer time and real-time thermal transfer speed, real-time slide vibration data, etc.) while controlling a thermal transfer device to print an image to be transferred on a slide surface;

[0097] The fault information of the thermal transfer device is determined based on the real-time printing parameters of multiple consecutive printing time points through a fault analysis model, wherein the fault analysis model can be a convolutional neural network (CNN) model.

[0098] In some embodiments, the quality detection module performs a quality assessment based on the image of the printed skateboard surface and the image features of the image to be transferred, including:

[0099] Extract the color distribution characteristics, detail level characteristics, gray-level co-occurrence matrix and autocorrelation function of the image of the printed skateboard surface;

[0100] Calculating image similarity between the image of the printed skateboard surface and the image to be transferred based on the color distribution characteristics, detail level characteristics, gray-level co-occurrence matrix, and autocorrelation function of the image of the printed skateboard surface and the color distribution characteristics, detail level characteristics, gray-level co-occurrence matrix, and autocorrelation function of the image to be transferred;

[0101] Quality assessment is performed based on image similarity.

[0102] Specifically, the method of calculating the image similarity between the printed skateboard surface image and the image to be transferred is similar to the method of calculating the image feature similarity, which will not be described in detail here.

[0103] Figure 3 This is a flow chart of a method for automatically generating printed graphics on a sports skateboard according to an embodiment of the present application. Figure 3 As shown, a method for automatically generating printed graphics and text on a sports skateboard may include the following process.

[0104] Step 310: establishing an image database, wherein the image database is used to store multiple template images;

[0105] Step 320 , obtaining user image and text requirements, and generating an image to be transferred based on the user image and text requirements and an image database;

[0106] Step 330 , obtaining relevant information of the sports skateboard and extracting image features of the image to be transferred;

[0107] Step 340: determining thermal transfer parameters based on the relevant information of the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters include at least thermal transfer temperature, thermal transfer pressure, thermal transfer time, and thermal transfer speed;

[0108] Step 350 , based on the thermal transfer parameters of the image to be transferred, controlling the thermal transfer device to print the image to be transferred on the surface of the slide;

[0109] Step 360 , obtaining an image of the printed surface of the skateboard, and performing a quality assessment based on the image of the printed surface of the skateboard and the image features of the image to be transferred.

[0110] A method for automatically generating printed text and graphics on a sports skateboard surface can be executed by a system for automatically generating printed text and graphics on a sports skateboard surface. For more descriptions of a method for automatically generating printed text and graphics on a sports skateboard surface, please refer to the relevant descriptions of a system for automatically generating printed text and graphics on a sports skateboard surface, which will not be repeated here.

[0111] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. A system for automatically generating printed graphics and text on a sports skateboard, characterized in that: include: An image management module, configured to establish an image database, wherein the image database is configured to store a plurality of template images; An image and text generation module, configured to obtain user image and text requirements, and generate an image to be transferred based on the user image and text requirements and the image database; a parameter determination module, configured to obtain relevant information about the sports skateboard, extract image features of the image to be transferred, and determine thermal transfer parameters based on the relevant information about the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters include at least thermal transfer temperature, thermal transfer pressure, thermal transfer time, and thermal transfer speed, and the relevant information about the sports skateboard includes material, size, and characteristics of the thermal transfer film; An image transfer module, configured to control a thermal transfer device to print the image to be transferred on the surface of the slide based on the thermal transfer parameters of the image to be transferred; a quality detection module, configured to obtain an image of the surface of the printed skateboard and perform a quality assessment based on the image of the surface of the printed skateboard and image features of the image to be transferred; The parameter determination module extracts image features of the image to be transferred, including: Obtaining adjacent pixel matrix characteristic parameters of each pixel of the image to be transferred, and determining edge pixels based on the adjacent pixel matrix characteristic parameters of each pixel; For the non-edge pixels of the image to be transferred, clustering the non-edge pixels of the image to be transferred based on the RGB value of each non-edge pixel to determine a plurality of non-edge pixel clusters; determining a color distribution feature of the image to be transferred based on the plurality of non-edge pixel clusters; determining detail level features of the image to be transferred based on the edge pixels; converting the image to be transferred into a grayscale image to be transferred; The grayscale co-occurrence matrix and autocorrelation function of the grayscale image to be transferred are extracted.

2. The automatic generation system of printed graphics and texts on a sports skateboard according to claim 1, characterized in that: The image management module establishes an image database, including: For each template image, extracting an image label of the template image; For any two template images, calculating the label similarity of the image labels of the two template images; Clustering the multiple template images according to any two of the template images to determine multiple image clusters; The image database is established based on the multiple image clusters.

3. The automatic generation system of printed graphics and texts on a sports skateboard according to claim 2, characterized in that: The image management module extracts the image tag of the template image, including: The image label of the template image is extracted through a label generation model, wherein the image label of the template image at least includes an object label, a scene label, an event label, an emotion label and a color feature label.

4. The automatic generation system of printed graphics and texts on a sports skateboard according to claim 2, characterized in that: The image and text generation module generates an image to be transferred based on the user's image and text requirements and the image database, including: Perform semantic understanding on the user's image and text requirements to determine image requirement information and text requirement information; Based on the image requirement information, determining a target template image from the image database; generating an intermediate image based on the image requirement information and the target template image through an image generation model; The image to be transferred is generated based on the text requirement information and the intermediate image through the image-text fusion model.

5. The automatic generation system of printed graphics and texts on a sports skateboard according to claim 4, characterized in that: The image-text generation module determines a target template image from the image database based on the image requirement information, including: determining a cluster matching value of each of the image clusters based on the image requirement keywords and the image tags of the template images included in each of the image clusters; determining a target image cluster from the plurality of image clusters based on the cluster matching degree of each of the image clusters; determining an image matching value of each template image included in the target image cluster based on the image requirement keyword and the image label of each template image included in the target image cluster; A target template image is determined from the target image cluster based on the image matching value of each template image included in the target image cluster.

6. The system for automatically generating printed graphics and text on a sports skateboard according to claim 1, characterized in that: The parameter determination module is further configured to: Establishing a processing sample database, wherein the processing sample database is used to store a plurality of processing samples, wherein the processing samples include image features of sample transfer images and sample thermal transfer parameters; The parameter determination module determines the thermal transfer parameters based on the relevant information of the moving skateboard and the image features of the image to be transferred, including: determining a target processing sample from the processing sample database based on image features of the image to be transferred; The thermal transfer parameters are determined based on the target processing sample and the image features of the image to be transferred through a parameter generation model.

7. The automatic generation system of printed graphics and texts on a sports skateboard according to claim 1, characterized in that: The quality detection module is also used for: acquiring real-time printing parameters during the process of controlling the thermal transfer device to print the image to be transferred on the surface of the skateboard; Fault information of the thermal transfer device is determined based on real-time printing parameters of a plurality of consecutive printing time points through a fault analysis model.

8. The system for automatically generating printed graphics and text on a sports skateboard according to claim 1, characterized in that: The quality detection module performs quality assessment based on the image of the printed skateboard surface and the image features of the image to be transferred, including: Extracting color distribution features, detail level features, gray-level co-occurrence matrix, and autocorrelation function of the image of the printed skateboard surface; Calculating image similarity between the printed skateboard surface image and the image to be transferred based on the color distribution characteristics, detail level characteristics, gray level co-occurrence matrix, and autocorrelation function of the printed skateboard surface image and the color distribution characteristics, detail level characteristics, gray level co-occurrence matrix, and autocorrelation function of the image to be transferred; Based on the image similarity, quality assessment is performed.

9. A method for automatically generating printed graphics on a sports skateboard, characterized in that: include: Establishing an image database, wherein the image database is used to store multiple template images; Acquiring a user's image and text requirements, and generating an image to be transferred based on the user's image and text requirements and the image database; Obtaining relevant information of the sports skateboard and extracting image features of the image to be transferred, wherein the relevant information of the sports skateboard includes material, size, and characteristics of the thermal transfer film; Determining thermal transfer parameters based on the relevant information of the sports skateboard and the image features of the image to be transferred, wherein the thermal transfer parameters at least include thermal transfer temperature, thermal transfer pressure, thermal transfer time, and thermal transfer speed; Based on the thermal transfer parameters of the image to be transferred, controlling the thermal transfer device to print the image to be transferred on the surface of the skateboard; Acquiring an image of the surface of the printed skateboard, and performing a quality assessment based on the image of the surface of the printed skateboard and image features of the image to be transferred; Extracting image features of the image to be transferred includes: Obtaining adjacent pixel matrix characteristic parameters of each pixel of the image to be transferred, and determining edge pixels based on the adjacent pixel matrix characteristic parameters of each pixel; For the non-edge pixels of the image to be transferred, clustering the non-edge pixels of the image to be transferred based on the RGB value of each non-edge pixel to determine a plurality of non-edge pixel clusters; determining a color distribution feature of the image to be transferred based on the plurality of non-edge pixel clusters; determining detail level features of the image to be transferred based on the edge pixels; converting the image to be transferred into a grayscale image to be transferred; The grayscale co-occurrence matrix and autocorrelation function of the grayscale image to be transferred are extracted.

Images