E-commerce image dynamic optimization and adjustment method based on market feedback

Abstract

The invention relates to the technical field of image processing, and particularly discloses an e-commerce image dynamic optimization adjustment method based on market feedback, which removes color noise and inhibits interference in edge detection through gray level conversion and smooth filtering technologies, extracts and analyzes edge curvature features by using Canny edge detection and wavelet transform, accurately identifies a sawtooth region, and improves the image quality. And performing boundary alignment on the pictures of the same commodity under different perspectives by using a shape context descriptor and a Hungary algorithm, calculating a shape difference coefficient to evaluate the consistency, and for further optimization processing, converting the shape difference coefficient and an edge curvature feature into a comprehensive feature vector as the input of a graph convolutional network model, so as to obtain the image convolutional network model. Global features are extracted through multi-layer graph convolution operation, an area with serious jagging is recognized, a hard threshold method is adopted for denoising, and the jagging area is accurately repaired through a cubic spline interpolation method.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to an e-commerce image dynamic optimization and adjustment method based on market feedback. Background Art

[0002] With the popularity of online shopping, consumers increasingly rely on high-quality product images to make purchasing decisions. However, due to shooting conditions, equipment differences, and improper post-processing, many product images have problems such as jagged, blurry, or inconsistent, which seriously affect consumers' shopping experience and trust in the products.

[0003] The prior art has the following deficiencies: In the e-commerce images fed back by the market, due to the complex shapes, irregular edges and diversity of multi-view displays in product images, jagged edges, discontinuities or positioning deviations often occur during edge detection. These problems will seriously affect the subsequent image optimization effects, such as background replacement and text layout accuracy, thereby reducing the attractiveness of product images and user conversion rates. Summary of the invention

[0004] The purpose of the present invention is to provide an e-commerce image dynamic optimization and adjustment method based on market feedback to solve the above-mentioned background problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions: The method for dynamic optimization and adjustment of e-commerce images based on market feedback includes the following steps: S1: Preprocess the product images and convert them into grayscale images to remove color noise that interferes with edge detection; S2: Extract edge curvature features from product images, calculate edge curvature changes, calibrate jagged areas based on edge curvature change thresholds, and identify specific locations of jagged edges. S3: Based on the jagged area, the images of the same product from multiple perspectives are aligned, the difference in boundary shape between different perspectives is calculated, and the consistency of the product images is evaluated based on the degree of difference; S4: The shape difference coefficient and edge curvature features of the product image are converted into a comprehensive feature vector as the input of the machine learning model, and the area with severe jaggedness is identified according to the model output results; S5: Repair edge curves in areas with severe jaggedness.

[0006] As a further solution of the present invention: the edge curvature feature is extracted as follows: The input product image is converted into a grayscale image, and Canny edge detection is used to obtain the set of all edge points in the image; Connect all edge points into a continuous edge curve and perform wavelet transform on the edge curve; Through wavelet transform, the curve is converted from time domain to frequency domain, so as to obtain the multi-scale features of the edge curve, which are recorded as edge curvature features.

[0007] As a further solution of the present invention: calculating the edge curvature according to the edge curvature feature, specifically including: According to the multi-scale wavelet coefficients of the curve, the curvature is calculated using the following formula: ; in, represents the rate of change of wavelet coefficients, represents the scale parameter, represents the translation parameter, represents the wavelet transform coefficient in the x-direction, Represents the wavelet transform coefficient in the y-direction; By calculating the derivative of the wavelet coefficients, the edge curvature of each edge point is obtained.

[0008] As a further solution of the present invention, according to the edge curvature of each edge point, the specific position of the edge jaggedness is identified, which specifically includes: Obtain the edge curvature of each edge point, compare the edge curvature of each edge point with a preset threshold, and determine whether the edge curvature of each edge point is greater than or equal to the preset threshold. If so, mark it as a jagged area, otherwise, mark it as a non-jagged area.

[0009] As a further solution of the present invention: the calculation process of the difference in boundary shape is: Get product images of the same product from multiple perspectives , Indicates the number of viewing angles and obtains the edge curve , Represents the boundary Points, Indicates the total number of sampling points; For each point in the boundary point set , construct shape context descriptor , the calculation expression is: ; in, Indicates sampling points, Indicate point and Point The Euclidean distance between and represents a point in the boundary point set, Indicate point and Point The polar angle, represents a discrete partition index, Represents the distance interval, represents the angle interval, Represents an indicator function, the value is 1 when the condition is met, otherwise it is 0. Indicate point Shape context descriptor of; For any two-view images and The boundary point set and , calculate the shape context descriptor of each point separately and , construct the cost matrix, each matrix element Indicate point With point The shape context difference is calculated as: ; In the formula, represents the number of distance partitions, Indicates the number of angular divisions, represents small positive values ​​that avoid zero denominators, Represents each matrix element; Use the Hungarian algorithm to solve the optimal matching of the cost matrix and establish the boundary point set and According to the one-to-one correspondence between the two boundary curves, the overall shape difference coefficient of the two boundary curves is calculated according to the point set matching results. The calculation expression is: ; In the formula, Represents the shape difference coefficient.

[0010] As a further solution of the present invention: evaluating the consistency of the product image specifically includes: It is determined whether the difference coefficient of the product image is greater than or equal to a preset threshold. If so, the images of the corresponding product under multiple viewing angles are inconsistent. If not, the images of the corresponding product under multiple viewing angles are consistent.

[0011] As a further solution of the present invention: the construction process of the comprehensive feature vector is: Obtaining shape difference coefficients and edge curvature features of product images, and standardizing the shape difference coefficients and edge curvature features; clustering the standardized shape difference coefficients and edge curvature feature vectors using a clustering algorithm to obtain a number of clusters, each cluster corresponding to a type of product image; calculating the center point of each cluster, the center point of the cluster being used as a comprehensive feature vector of the product images in the corresponding cluster, the comprehensive feature vector representing the common features of the images in the cluster; assigning each product image to a corresponding cluster, and using the center point of the corresponding cluster as the comprehensive feature vector of the product image; The clustering algorithm is a K-means clustering algorithm, and the specific process is: randomly select M initial cluster centers; calculate the distance between the feature vector of each image and all center points, and assign the image to the cluster center with the nearest distance; update the center point of each cluster to the mean of the feature vectors of all images in the cluster; repeat the clustering process until the cluster center no longer changes, and the clustering process ends.

[0012] As a further solution of the present invention: the construction process of the machine learning model is: The shape difference coefficient and edge curvature features of the product image are replaced with a comprehensive feature vector as the input item of the machine learning model, and the machine learning model is a graph neural network model; the graph convolution operation is used to gradually extract local and global features, and finally a global graph representation is generated; through the processing of multi-layer graph convolution, the graph neural network can recognize each set of input comprehensive feature vectors, adjust the model parameters by minimizing the error between the predicted results and the actual annotations, and train the machine learning model until the sum of the prediction errors reaches convergence. The model training is stopped, and finally an jagged degree score is generated to identify areas with severe jaggedness.

[0013] As a further solution of the present invention: identifying a severe jagged area specifically includes: According to the recognition result of the machine learning model, a jaggedness score is output for each jagged area; It is determined whether the jaggedness score of each jagged region is greater than or equal to a preset threshold. If so, the corresponding region is a severely jagged region; if not, the corresponding region is a non-severely jagged region.

[0014] As a further solution of the present invention: for the area with severe jaggedness, the edge curve is repaired, which specifically includes: Use hard thresholding method to denoise the wavelet coefficients and remove high-frequency noise in the image; The wavelet coefficients after denoising and The cubic spline interpolation method is used to repair the local curve and calculate the curve fitting function of each jagged area. The calculation expression is: ; in, represents the arc length parameter of the curve, represents the curve fitting function, represents a spline, represents the total number of splines, Indicates The fitting coefficients of the splines, Indicates Spline basis functions; Get the edge curvature of the jagged area , optimize the edge curvature, use the least squares method to adjust the fitting curve, and obtain the anti-aliased image.

[0015] Beneficial effects of the present invention: (1) Through a series of advanced image processing technologies, including grayscale conversion, edge detection, wavelet transform, and the application of shape context descriptors, the present invention can accurately identify and repair jagged areas in product images, thereby improving the quality and visual effect of a single image. In addition, by aligning the boundaries of images of the same product from multiple perspectives and calculating the shape difference coefficient to evaluate their consistency, the high consistency and coordination of multi-perspective display is ensured. This high-quality and consistent product display not only enhances the visual appeal of the product, but also effectively reduces consumers' concerns about image quality issues and enhances user trust. More importantly, by optimizing the overall presentation of product images, the user's shopping experience and satisfaction can be significantly improved, thereby increasing sales conversion rates. Specifically, the present invention uses a graph neural network model to extract global and local features from a comprehensive feature vector, combined with advanced image restoration techniques (such as cubic spline interpolation), to achieve accurate restoration of severely jagged areas, thereby providing an efficient, automated, and highly practical product image optimization solution for e-commerce platforms.

[0016] (2) The present invention proposes a fully automated image optimization method, which covers all aspects from preprocessing, feature extraction, consistency evaluation to image restoration, and is used for efficient processing of large-scale product images on e-commerce platforms. By adopting advanced grayscale conversion, edge detection and wavelet transform techniques for preprocessing, and using shape context descriptors to accurately identify jagged areas, the method ensures high-quality single image output. Furthermore, a machine learning model based on a graph convolutional network (GCN) performs boundary alignment and consistency evaluation on product images from multiple perspectives, and can intelligently identify jagged areas that need to be repaired. Combining optimization algorithms (such as the Hungarian algorithm) and automated restoration techniques (such as cubic spline interpolation), the method achieves accurate restoration of jagged areas, greatly reduces the need for manual intervention, and significantly improves processing efficiency and the stability of results. This highly automated process not only saves time and cost, but also enables e-commerce platforms to batch update product images in a short period of time. In addition, by improving the quality and consistency of product display, the present invention also effectively enhances user experience and trust. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below in conjunction with the accompanying drawings.

[0018] Figure 1 It is a flowchart of the specific steps of the method for dynamic optimization and adjustment of e-commerce images based on market feedback of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] See also Figure 1 As shown, the present invention is a method for dynamic optimization and adjustment of e-commerce images based on market feedback, comprising the following steps: S1: Preprocess the product images and convert them into grayscale images to remove color noise that interferes with edge detection; S2: Extract edge curvature features from product images, calculate edge curvature changes, calibrate jagged areas based on edge curvature change thresholds, and identify specific locations of jagged edges. S3: Based on the jagged area, the images of the same product from multiple perspectives are aligned, the difference in boundary shape between different perspectives is calculated, and the consistency of the product images is evaluated based on the degree of difference; S4: The shape difference coefficient and edge curvature features of the product image are converted into a comprehensive feature vector as the input of the machine learning model, and the area with severe jaggedness is identified according to the model output results; S5: Repair edge curves in areas with severe jaggedness.

[0021] In S1, the product image is preprocessed and converted into a grayscale image to remove color noise that interferes with edge detection, including: In the image preprocessing process of product images, the color images are first converted from the RGB color space to grayscale images. This operation is completed by weighted averaging the red, green, and blue components of each pixel to generate a single grayscale value, thereby retaining the brightness information of the image and eliminating the interference of color differences on subsequent processing. This conversion not only simplifies data processing, but also improves the computational efficiency of the image, providing a purer image basis for subsequent edge detection.

[0022] Next, the converted grayscale image is smoothed to suppress the influence of noise. The filtering method is used to remove the fine noise in the grayscale image, smooth the edges and avoid the influence of interference information on edge detection. This process can effectively reduce the interference of random noise and detail information on subsequent analysis, ensure that the extracted edge curvature features are more accurate and stable, and lay a reliable foundation for the subsequent recognition of jagged areas.

[0023] In S2, the edge curvature features in the product image are extracted, and the edge curvature change is calculated. The jagged area is calibrated according to the threshold of the edge curvature change, and the specific location of the edge jagged is identified, which includes: The edge curvature feature extraction is as follows: The input product image is converted into a grayscale image, and Canny edge detection is used to obtain the set of all edge points in the image; Connect all edge points into a continuous edge curve and perform wavelet transform on the edge curve; Through wavelet transform, the curve is converted from time domain to frequency domain, so as to obtain the multi-scale features of the edge curve, which are recorded as edge curvature features.

[0024] According to the edge curvature characteristics, the edge curvature is calculated, including: According to the multi-scale wavelet coefficients of the curve, the curvature is calculated using the following formula: ; in, represents the rate of change of wavelet coefficients, represents the scale parameter, represents the translation parameter, represents the wavelet transform coefficient in the x-direction, Represents the wavelet transform coefficient in the y-direction; By calculating the derivative of the wavelet coefficients, the edge curvature of each edge point is obtained; According to the edge curvature of each edge point, the specific location of the edge jaggedness is identified, including: Obtain the edge curvature of each edge point, compare the edge curvature of each edge point with a preset threshold, and determine whether the edge curvature of each edge point is greater than or equal to the preset threshold. If so, mark it as a jagged area, otherwise, mark it as a non-jagged area.

[0025] It should be noted that: by using wavelet transform technology to perform multi-scale analysis on edge curves in product images, the edge curvature features can be accurately extracted, and the position of the jagged area can be quickly calibrated in combination with the edge curvature threshold; wavelet transform is the core step in extracting edge curvature features, and the type of wavelet function (such as Haar wavelet) and parameter settings (such as scale parameters and translation parameters) used.

[0026] In S3, based on the jagged area, the borders of the images of the same product from multiple perspectives are aligned, the difference in the border shapes between different perspectives is calculated, and the consistency of the product images is evaluated based on the degree of difference, including: According to the difference degree of boundary shape between different viewing angles, the shape difference coefficient is calculated to evaluate the consistency of the product image; the calculation process of the shape difference coefficient is as follows: Get product images of the same product from multiple perspectives , Indicates the number of viewing angles and obtains the edge curve , Represents the boundary Points, Indicates the total number of sampling points; For each point in the boundary point set , construct shape context descriptor , the calculation expression is: ; in, Indicates sampling points, Indicate point and Point The Euclidean distance between and represents a point in the boundary point set, Indicate point and Point The polar angle, represents a discrete partition index, Represents the distance interval, represents the angle interval, Represents an indicator function, the value is 1 when the condition is met, otherwise it is 0. Indicate point Shape context descriptor of; For any two-view images and The boundary point set and , calculate the shape context descriptor of each point separately and , construct the cost matrix, each matrix element Indicate point With point The shape context difference is calculated as: ; In the formula, represents the number of distance partitions, Indicates the number of angular divisions, represents small positive values ​​that avoid zero denominators, Represents each matrix element; Use the Hungarian algorithm to solve the optimal matching of the cost matrix and establish the boundary point set and According to the one-to-one correspondence between the two boundary curves, the overall shape difference coefficient of the two boundary curves is calculated according to the point set matching results. The calculation expression is: ; In the formula, represents the shape difference coefficient; Determine whether the difference coefficient of the product image is greater than or equal to a preset threshold. If so, the images of the corresponding product under multiple viewing angles are inconsistent. If not, the images of the corresponding product under multiple viewing angles are consistent. It should be noted that the shape difference coefficient is calculated to evaluate the consistency of images of the same product under different viewing angles, and the shape context descriptor and the Hungarian algorithm are used to achieve the optimal matching of the boundary point set, so as to effectively identify the inconsistency between images. It can not only quantify the similarity of the boundaries of product images under different viewing angles, but also effectively identify the inconsistency caused by the jagged areas, thereby ensuring the accurate evaluation of the consistency of product images.

[0027] In S4, the shape difference coefficient and edge curvature features of the product image are converted into a comprehensive feature vector as the input of the machine learning model. The areas with severe jaggedness are identified according to the model output results, including: Obtaining shape difference coefficients and edge curvature features of product images, and standardizing the shape difference coefficients and edge curvature features; clustering the standardized shape difference coefficients and edge curvature feature vectors using a clustering algorithm to obtain a number of clusters, each cluster corresponding to a type of product image; calculating the center point of each cluster, the center point of the cluster being used as a comprehensive feature vector of the product images in the corresponding cluster, the comprehensive feature vector representing the common features of the images in the cluster; assigning each product image to a corresponding cluster, and using the center point of the corresponding cluster as the comprehensive feature vector of the product image; The clustering algorithm is a K-means clustering algorithm, and the specific process is: randomly select M initial cluster centers; calculate the distance between the feature vector of each image and all the center points, and assign the image to the cluster center with the closest distance; update the center point of each cluster to the mean of the feature vectors of all images in the cluster; repeat the clustering process until the cluster center no longer changes, and the clustering process ends; The shape difference coefficient and edge curvature features are extracted from product images at different viewing angles, and these features are combined into a comprehensive feature vector. The shape difference coefficient reflects the inconsistency of the boundary shape under different viewing angles, while the edge curvature feature captures the curvature degree and change trend of the boundary line.

[0028] Each product view is regarded as a node in the graph, and the comprehensive feature vector is used as the node attribute. If there is a significant shape or position relationship between two views, an edge is established between the two nodes, and the weight of the edge can be defined according to the similarity or distance between them.

[0029] The graph structure is processed using a graph convolutional network (GCN). Through multi-layer graph convolution operations, the model can gradually extract local and global feature information. Each layer of graph convolution enhances the node's understanding of the features of its neighboring nodes, thereby better capturing complex patterns and associations.

[0030] After multiple layers of graph convolution, a global graph representation is finally generated, which integrates the information of all nodes and their relationships. This global graph representation can effectively reflect the overall characteristics of the entire product image set.

[0031] The model is trained using annotated data sets, and the model parameters are adjusted by minimizing the error between the predicted results and the actual annotations. During the training process, an appropriate loss function (such as mean square error) is used to measure the gap between the predicted value and the true value, and the model parameters are updated through the back propagation algorithm until the sum of the prediction errors reaches the convergence standard.

[0032] Once the model is trained, it can be used to evaluate a new set of product images to generate an aliasing score that can help identify product images with severe aliasing areas and provide guidance for image optimization.

[0033] In S5, the edge curves of severely jagged areas are repaired, including: Use hard thresholding method to denoise the wavelet coefficients and remove high-frequency noise in the image; The wavelet coefficients after denoising and The cubic spline interpolation method is used to repair the local curve and calculate the curve fitting function of each jagged area. The calculation expression is: ; in, represents the arc length parameter of the curve, represents the curve fitting function, represents a spline, represents the total number of splines, Indicates The fitting coefficients of the splines, Indicates Spline basis functions; Get the edge curvature of the jagged area , optimize the edge curvature, use the least squares method to adjust the fitting curve, and obtain the anti-aliased image.

[0034] Working principle of the present invention: The present invention converts the original product image into a grayscale image to remove color noise, and smoothes it through filtering technology to ensure the accuracy of subsequent edge detection. The edge curvature features in the product image are extracted and analyzed, and the Canny edge detection and wavelet transform are used to identify the specific location of the edge jaggedness, so as to accurately calibrate the jagged area. Next, the borders of the images of the same product under different viewing angles are aligned, the shape difference coefficient is calculated to evaluate its consistency, and the shape context descriptor and the Hungarian algorithm are used to achieve the optimal match. The shape difference coefficient is combined with the edge curvature feature and converted into a comprehensive feature vector as the input of the machine learning model (such as the neural network in the figure), and the global features are extracted through multi-layer graph convolution operations, and the model is trained to identify the severe jagged area. Finally, the hard threshold method is used for denoising, and the cubic spline interpolation method is used to repair the edge curve with severe jaggedness to eliminate the jagged phenomenon in the image. This method not only improves the overall quality of the product image, but also improves the user browsing experience. It is particularly suitable for large-scale product image processing on e-commerce platforms, which helps to enhance the product display effect and improve the sales conversion rate. The entire process emphasizes comprehensive optimization from preprocessing, feature extraction, consistency assessment to image restoration, demonstrating a high degree of technological innovation and application value.

[0035] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0036] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). 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 contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0037] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0038] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0039] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for dynamic optimization and adjustment of e-commerce images based on market feedback, characterized in that: The following steps are involved: S1: Preprocess the product image and convert it into a grayscale image to remove color noise that interferes with edge detection; S2: Extract edge curvature features from product images, calculate edge curvature changes, calibrate jagged areas based on edge curvature change thresholds, and identify specific locations of jagged edges. S3: Based on the jagged area, the images of the same product from multiple perspectives are aligned, the difference in boundary shape between different perspectives is calculated, and the consistency of the product images is evaluated based on the degree of difference; S4: The shape difference coefficient and edge curvature features of the product image are converted into a comprehensive feature vector as the input of the machine learning model, and the area with severe jaggedness is identified according to the model output results; S5: Repair edge curves in areas with severe jaggedness.

2. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1 is characterized in that: The edge curvature feature extraction is as follows: The input product image is converted into a grayscale image, and Canny edge detection is used to obtain the set of all edge points in the image; Connect all edge points into a continuous edge curve and perform wavelet transform on the edge curve; Through wavelet transform, the curve is converted from time domain to frequency domain, so as to obtain the multi-scale features of the edge curve, which are recorded as edge curvature features.

3. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 2 is characterized in that: According to the edge curvature characteristics, the edge curvature is calculated, including: According to the multi-scale wavelet coefficients of the curve, the curvature is calculated using the following formula: ; in, represents the rate of change of wavelet coefficients, represents the scale parameter, represents the translation parameter, represents the wavelet transform coefficient in the x-direction, Represents the wavelet transform coefficient in the y-direction; By calculating the derivative of the wavelet coefficients, the edge curvature of each edge point is obtained.

4. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 3 is characterized in that: According to the edge curvature of each edge point, the specific location of the edge jaggedness is identified, including: Obtain the edge curvature of each edge point, compare the edge curvature of each edge point with a preset threshold, and determine whether the edge curvature of each edge point is greater than or equal to the preset threshold. If so, mark it as a jagged area, otherwise, mark it as a non-jagged area.

5. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1 is characterized in that: The difference in boundary shape is calculated as: Get product images of the same product from multiple perspectives , Indicates the number of viewing angles and obtains the edge curve , Represents the boundary Points, Indicates the total number of sampling points; For each point in the boundary point set , construct shape context descriptor , the calculation expression is: ; in, Indicates sampling points, Indicate point and Point The Euclidean distance between and represents a point in the boundary point set, Indicate point and Point The polar angle, represents a discrete partition index, represents the distance interval, represents the angle interval, Represents an indicator function, the value is 1 when the condition is met, otherwise it is 0. Indicate point Shape context descriptor of; For any two-view images and The boundary point set and , calculate the shape context descriptor of each point separately and , construct the cost matrix, each matrix element Indicate point With point The shape context difference is calculated as: ; In the formula, represents the number of distance partitions, Indicates the number of angular divisions, represents small positive values ​​that avoid zero denominators, Represents each matrix element; Use the Hungarian algorithm to solve the optimal matching of the cost matrix and establish the boundary point set and According to the one-to-one correspondence between the two boundary curves, the overall shape difference coefficient of the two boundary curves is calculated according to the point set matching results. The calculation expression is: ; In the formula, Represents the shape difference coefficient.

6. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1, characterized in that: Evaluating the consistency of product images specifically includes: It is determined whether the difference coefficient of the product image is greater than or equal to a preset threshold. If so, the images of the corresponding product under multiple viewing angles are inconsistent. If not, the images of the corresponding product under multiple viewing angles are consistent.

7. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1, characterized in that: The construction process of the comprehensive feature vector is: Obtaining shape difference coefficients and edge curvature features of product images, and standardizing the shape difference coefficients and edge curvature features; clustering the standardized shape difference coefficients and edge curvature feature vectors using a clustering algorithm to obtain a number of clusters, each cluster corresponding to a type of product image; calculating the center point of each cluster, the center point of the cluster being used as a comprehensive feature vector of the product images in the corresponding cluster, the comprehensive feature vector representing the common features of the images in the cluster; Assign each product image to a corresponding cluster, and use the center point of the corresponding cluster as the comprehensive feature vector of the product image; The clustering algorithm is a K-means clustering algorithm, and the specific process is: randomly select M initial cluster centers; calculate the distance between the feature vector of each image and all center points, and assign the image to the cluster center with the nearest distance; update the center point of each cluster to the mean of the feature vectors of all images in the cluster; repeat the clustering process until the cluster center no longer changes, and the clustering process ends.

8. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1, characterized in that: The process of building a machine learning model is: The shape difference coefficient and edge curvature features of the product image are replaced with a comprehensive feature vector as the input item of the machine learning model, and the machine learning model is a graph neural network model; the graph convolution operation is used to gradually extract local and global features, and finally a global graph representation is generated; through the processing of multi-layer graph convolution, the graph neural network can recognize each set of input comprehensive feature vectors, adjust the model parameters by minimizing the error between the predicted results and the actual annotations, and train the machine learning model until the sum of the prediction errors reaches convergence. The model training is stopped, and finally an jagged degree score is generated to identify areas with severe jaggedness.

9. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 8, characterized in that: Identify areas of severe aliasing, including: According to the recognition result of the machine learning model, a jaggedness score is output for each jagged area; It is determined whether the jaggedness score of each jagged region is greater than or equal to a preset threshold. If so, the corresponding region is a severely jagged region; if not, the corresponding region is a non-severely jagged region.

10. The method for dynamic optimization and adjustment of e-commerce images based on market feedback according to claim 1, characterized in that: For areas with severe jagged edges, the edge curves are repaired, including: Use hard thresholding method to denoise the wavelet coefficients and remove high-frequency noise in the image; The wavelet coefficients after denoising and The cubic spline interpolation method is used to repair the local curve and calculate the curve fitting function of each jagged area. The calculation expression is: ;in, represents the arc length parameter of the curve, represents the curve fitting function, represents a spline, represents the total number of splines, Indicates The fitting coefficients of the splines, Indicates Spline basis functions; Get the edge curvature of the jagged area , optimize the edge curvature, use the least squares method to adjust the fitting curve, and obtain the anti-aliased image.

Images