Virtual product digital image processing method and system

Abstract

The invention is suitable for the technical field of image processing, and provides a virtual product digital image processing method and system, and the method comprises the following steps: sampling a virtual product image, and carrying out the image coding and compression of the sampled virtual product image; performing image enhancement and 3D model rendering on the compressed virtual product image; identifying key features in the virtual product image, wherein the key features comprise limb features and expression features; generating a plurality of derivative virtual product images based on the limb features and the expression features, wherein each derivative virtual product image corresponds to one interaction entry; an icon button is generated for each interaction entry, and a user clicks the icon buttons to achieve dynamic interaction with the virtual product image; and receiving a voice binding instruction, binding a section of voice for each interaction entry, and synchronously playing the voice when the corresponding icon button is triggered. Therefore, the user can dynamically interact with the virtual product, and the interestingness and interactivity of the virtual product are increased.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and system for processing digital images of virtual products. Background Art

[0002] With the rapid development of information technology, virtual products are increasingly used in games, film and television, education, e-commerce and other fields, and virtual animation images have become an indispensable part of the digital world. Traditional virtual product image processing methods mainly focus on the static display of images, presenting the appearance of virtual products through simple two-dimensional images or three-dimensional models. However, this method has many shortcomings in terms of image clarity, detail expression and interactivity. Virtual products that lack dynamic interaction are difficult to attract users' continued attention and interest, and cannot meet users' needs for diversified presentation and personalized interaction of virtual products. Therefore, it is necessary to provide a virtual product digital image processing method and system to solve the above problems. Summary of the invention

[0003] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for processing virtual product digital images to solve the problems existing in the above-mentioned background technology.

[0004] The present invention is implemented as follows: a method for processing a virtual product digital image, the method comprising the following steps: Sampling virtual product images, and performing image encoding and compression on the sampled virtual product images; Performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; Identify key features in the virtual product image, wherein the key features include body features and facial features; Generate a number of derivative virtual product images based on body features and expression features, each derivative virtual product image corresponds to an interactive item; Generate an icon button for each interactive item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; Receive voice binding instructions, bind a voice to each interactive item, and play the voice synchronously when the corresponding icon button is triggered.

[0005] As a further solution of the present invention: the step of sampling the virtual product image and encoding and compressing the sampled virtual product image specifically includes: Convert the continuous virtual product image signal into discrete data points according to the set sampling rate; Quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; The quantized discrete levels are mapped back to the frequency domain coefficients, the frequency domain coefficients are converted back to the spatial domain to obtain the restored image, the restored image is denoised and the color space is converted to obtain the processed virtual product image.

[0006] As a further solution of the present invention: the step of performing image enhancement and 3D model rendering on the compressed virtual product image specifically includes: Sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of virtual product images; Obtain the geometric information of the model corresponding to the virtual product image, transform the model from the world coordinate system to the observation coordinate system, project the 3D model in the observation coordinate system onto a 2D plane, and generate a perspective projection or parallel projection image; Define the light source information, emit light from the camera position, enter the scene for tracking, calculate the intersection point of the light and the model in the scene, as well as the reflection and refraction interaction effects of the light on the model surface, and determine the shadow effect.

[0007] As a further solution of the present invention: the step of identifying key features in the virtual product image specifically includes: Perform limb feature recognition, use edge detection and contour extraction to identify limb parts in virtual product images, and use posture estimation algorithms to determine the position and posture of limbs; Perform facial expression feature recognition, use facial detection algorithm to locate the facial area, extract feature points in the facial area, and perform facial expression recognition based on the position and changes of feature points.

[0008] As a further solution of the present invention: the step of generating a plurality of derivative virtual product images based on the body features and expression features specifically includes: Retrieve the body movements and facial expressions corresponding to each interaction item; The recognized body features and facial features are integrated with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction items; The derived virtual product image is rendered and detail adjusted.

[0009] Another object of the present invention is to provide a virtual product digital image processing system, the system comprising: A virtual image compression module, used for sampling virtual product images, and performing image encoding and compression on the sampled virtual product images; A virtual image rendering module, used for performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; A key feature recognition module, used to recognize key features in a virtual product image, wherein the key features include body features and facial features; A derived virtual image module, used to generate a number of derived virtual product images based on body features and expression features, each derived virtual product image corresponding to an interactive entry; An icon button interaction module is used to generate an icon button for each interaction item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; The voice information binding module is used to receive voice binding instructions, bind a voice segment to each interactive item, and play the voice synchronously when the corresponding icon button is triggered.

[0010] As a further solution of the present invention: the virtual image compression module includes: An image signal conversion unit, used for converting a continuous virtual product image signal into discrete data points according to a set sampling rate; An image data quantization unit, used for quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; The image decoding and restoration unit is used to map the quantized discrete levels back to the frequency domain coefficients, convert the frequency domain coefficients back to the spatial domain to obtain a restored image, perform denoising and color space conversion on the restored image, and obtain a processed virtual product image.

[0011] As a further solution of the present invention: the virtual image rendering module includes: A virtual image enhancement unit, used for sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of virtual product images; A geometric information transformation unit is used to obtain geometric information of a model corresponding to a virtual product image, transform the model from a world coordinate system to an observation coordinate system, project the 3D model in the observation coordinate system onto a 2D plane, and generate an image of perspective projection or parallel projection; The model ray tracing unit is used to define the light source information, emit light from the camera position, enter the scene for tracing, calculate the intersection points of the light and the model in the scene, as well as the reflection and refraction interaction effects of the light on the model surface, and determine the shadow effect.

[0012] As a further solution of the present invention: the key feature recognition module includes: A limb feature recognition unit is used to perform limb feature recognition, identify limb parts in the virtual product image using edge detection and contour extraction, and determine the position and posture of the limb using a posture estimation algorithm; The expression feature recognition unit is used to perform expression feature recognition, locate the facial area using a facial detection algorithm, extract feature points within the facial area, and perform expression recognition based on the position and changes of the feature points.

[0013] As a further solution of the present invention: the derived virtual image module includes: An information retrieval unit, used to retrieve the body movements and facial expressions corresponding to each interaction item; A derivative image determination unit, used to fuse the recognized body features and expression features with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction items; The derivative image processing unit is used to render and adjust the details of the derivative virtual product image.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention significantly improves the clarity and detail of virtual product images by sampling virtual product images and performing image encoding and compression, as well as subsequent image enhancement and 3D model rendering. The application of geometric calculation and ray tracing technology further enhances the three-dimensional sense and realism of virtual products, providing users with a more realistic visual experience. By identifying key features in virtual product images (such as body features and expression features), several derivative virtual product images can be generated, each of which corresponds to an interactive item. Users can achieve dynamic interaction with virtual products by clicking icon buttons, thereby increasing the fun and interactivity of virtual products. The present invention allows binding a voice to each interactive item, and when the user triggers the corresponding icon button, the voice is played synchronously. This synchronous interaction between voice and image not only enriches the user's interactive experience, but also meets the user's needs for personalized presentation and interaction of virtual products. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flowchart of a method for processing digital images of virtual products.

[0016] Figure 2 The present invention is a flowchart of image encoding and compression in a virtual product digital image processing method.

[0017] Figure 3 A flowchart of image enhancement and 3D model rendering in a virtual product digital image processing method.

[0018] Figure 4 Flowchart for identifying key features in a virtual product digitization image processing method.

[0019] Figure 5The invention discloses a flow chart of generating a plurality of derivative virtual product images in a virtual product digitized image processing method.

[0020] Figure 6 This is a structural diagram of a virtual product digital image processing system. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0023] like Figure 1 As shown, an embodiment of the present invention provides a method for processing a virtual product digital image, the method comprising the following steps: S100, sampling a virtual product image, and performing image encoding and compression on the sampled virtual product image; S200, performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; S300, identifying key features in the virtual product image, wherein the key features include body features and facial expression features; S400, generating a plurality of derivative virtual product images based on the body features and the expression features, each derivative virtual product image corresponding to an interactive item; S500, generating an icon button for each interactive item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; S600, receiving a voice binding instruction, binding a voice segment to each interactive item, and playing the voice synchronously when the corresponding icon button is triggered.

[0024] It should be noted that the traditional virtual product image processing method mainly focuses on the static display of images, presenting the appearance of virtual products through simple two-dimensional images or three-dimensional models. However, this method has many deficiencies in image clarity, detail expression and interactivity. Virtual products lacking dynamic interaction are difficult to attract users' continuous attention and interest, and cannot meet users' needs for diversified presentation and personalized interaction of virtual products. The embodiments of the present invention are intended to solve the above problems.

[0025] In the embodiment of the present invention, firstly, a virtual product image is sampled, and the sampled virtual product image is image encoded and compressed. It is easy to understand that in order to improve efficiency in network transmission and storage, image data needs to be compressed, and the selection of compression algorithm should take into account image quality, compression ratio and processing speed. Then, the compressed virtual product image is image enhanced and 3D model rendered, and the 3D model rendering includes geometric calculation and ray tracing to generate a realistic three-dimensional image with more prominent details. Then, the key features in the virtual product image are identified, and the key features include body features and expression features, and a number of derivative virtual product images are generated according to the body features and expression features. Each derivative virtual product image corresponds to an interactive item, such as greeting, happiness, anger, etc., and an icon button is generated for each interactive item. A series of icon buttons are set at the bottom of the virtual product image. In this way, the user clicks the icon button to realize dynamic interaction with the virtual product image, meeting the user's diversified and personalized interaction needs; in addition, the user can also enter a voice binding instruction to bind a voice to each interactive item. When the corresponding icon button is triggered, the voice is played synchronously, and the interaction will be more interesting.

[0026] like Figure 2 As shown, as a preferred embodiment of the present invention, the step of sampling the virtual product image and encoding and compressing the sampled virtual product image specifically includes: S101, converting continuous virtual product image signals into discrete data points according to a set sampling rate; S102, quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; S103, mapping the quantized discrete levels back to frequency domain coefficients, converting the frequency domain coefficients back to the spatial domain to obtain a restored image, performing denoising and color space conversion on the restored image to obtain a processed virtual product image.

[0027] In the embodiment of the present invention, the continuous virtual product image signal is first converted into discrete data points according to the set sampling rate, and then the converted data is quantized into discrete grayscale values ​​or color values. The higher the quantization level, the richer the color and grayscale representation of the image, but the amount of data will also increase accordingly, and the quantized data is encoded and compressed into a digital signal. Finally, decoding is performed, and decoding is the inverse process of encoding, specifically including: mapping the quantized discrete level back to the frequency domain coefficient, converting the frequency domain coefficient back to the spatial domain, obtaining the restored image, denoising and color space conversion of the restored image, and obtaining the processed virtual product image.

[0028] As a preferred embodiment of the present invention, mapping the quantized discrete levels back to frequency domain coefficients, converting the frequency domain coefficients back to the spatial domain, and obtaining a restored image specifically includes the following sub-steps: The quantized discrete levels are compressed using discrete cosine transform to generate a frequency domain coefficient matrix; According to the position of each element in the frequency domain coefficient matrix, a corresponding first adaptive weight is adaptively given, and a weighted calculation is performed to obtain a weighted frequency domain coefficient matrix; Perform nonlinear mapping on the absolute value of each element in the weighted frequency domain coefficient matrix to obtain an enhanced frequency domain coefficient matrix; Counting the energy of the frequency components of the frequency domain coefficient matrix in each direction after the enhancement, and determining the enhancement direction according to the energy distribution; According to the position of each frequency component, calculate the projection value of the frequency component in the current direction; Perform Gaussian distribution processing on the projection values ​​to obtain the directional enhancement weight matrix; Applying the directional enhancement weight matrix to the enhanced frequency domain coefficient matrix to obtain a directional enhanced frequency domain coefficient matrix; The frequency domain coefficient matrix of the direction enhancement is subjected to two-dimensional inverse discrete cosine transform to obtain the restored image.

[0029] In an embodiment of the present invention, by introducing adaptive weights and adjusting the contribution of each coefficient according to the importance of the frequency component, the influence of high-frequency noise can be effectively reduced, while retaining the image structure information that dominates the low-frequency components, thereby ensuring the quality of image reconstruction. Nonlinear enhancement of frequency domain coefficients can highlight important high-intensity components (such as edges and textures of the image) while suppressing noise that may exist in low-intensity components. This processing can enhance image details, making virtual product images more realistic and attractive. Moreover, by adaptively selecting directions for enhancement, the clarity of edges and textures can be significantly improved. In particular, in virtual product images, this enhancement method can make the contours, textures, and surface details of the product more prominent, thereby attracting user attention.

[0030] As a preferred embodiment of the present invention, the denoising and color space conversion of the restored image to obtain the processed virtual product image specifically includes the following sub-steps: Perform multi-scale decomposition on the restored image to decompose the image into multiple levels to obtain high-frequency detail features and low-frequency smooth features; Perform gradient calculation on the restored image to obtain the gradient amplitude of the image; Normalize the gradient amplitude of the image to obtain a gradient normalization result; Adaptively generating a second adaptive weight according to the gradient amplitude in the gradient normalization result; Applying multi-scale decomposition to the restored image to obtain multi-scale detail features, wherein the multi-scale detail features include high-frequency detail features and low-frequency smooth features; The noise suppression algorithm is used to denoise the high-frequency detail features, and the fuzzy processing is used to denoise the low-frequency smooth features to obtain multi-scale denoising results; Combining the multi-scale denoising results according to the second adaptive weight to obtain a denoised image; The denoised image is converted into a color space to obtain a processed virtual product image.

[0031] In the embodiment of the present invention, based on the weight distribution of local gradients, denoising strategies of different strengths can be adopted in different areas. For example, areas with complex textures retain more details, while smooth areas are strongly denoised. This targeted processing greatly improves the naturalness and delicacy of the image. In addition, by processing the detail layer and the smooth layer at a multi-scale decomposition, adaptive denoising of local areas can be achieved, which can both smooth the noise and retain key high-frequency details (such as edges and textures). This is especially important for virtual product images, because the material, texture and delicate design of the product need to be preserved. Figure 3 As shown, as a preferred embodiment of the present invention, the step of performing image enhancement and 3D model rendering on the compressed virtual product image specifically includes: S201, sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of the virtual product image; S202, obtaining geometric information of a model corresponding to the virtual product image, converting the model from a world coordinate system to an observation coordinate system, projecting the 3D model in the observation coordinate system onto a 2D plane, and generating an image of perspective projection or parallel projection; S203, define light source information, emit light from the camera position, enter the scene for tracking, calculate the intersection point of the light and the model in the scene, and the reflection and refraction interaction effects of the light on the surface of the model, and determine the shadow effect.

[0032] In the embodiment of the present invention, before the 3D model is rendered, the virtual product image needs to be sharpened, contrast adjusted, color corrected, color saturation adjusted, and spatial domain enhanced to enhance the edges and details of the image, make the image clearer, adjust the color balance of the image to make it more natural, increase or decrease the color saturation of the image to make it more vivid or soft, and use various filters to smooth, sharpen, or detect edges on the image. Then, the geometric information of the model corresponding to the virtual product image is retrieved, including the vertices, edges, and faces of the model, and the model is converted from the world coordinate system to the observation coordinate system. The 3D model in the observation coordinate system is projected onto a 2D plane to generate a perspective projection or parallel projection image. Finally, it is necessary to define the light source information (light source type, intensity, color, and other parameters), emit light from the camera position, enter the scene for tracking, calculate the intersection point of the light and the model in the scene, and the reflection and refraction interaction effects of the light on the surface of the model, determine the shadow effect, and thus generate a highly realistic image.

[0033] like Figure 4 As shown, as a preferred embodiment of the present invention, the step of identifying key features in the virtual product image specifically includes: S301, performing limb feature recognition, using edge detection and contour extraction to identify limb parts in the virtual product image, and using a posture estimation algorithm to determine the position and posture of the limb; S302, performing facial expression feature recognition, using a facial detection algorithm to locate the facial area, extracting feature points in the facial area, and performing facial expression recognition based on the positions and changes of the feature points.

[0034] In the embodiment of the present invention, when performing body feature recognition, edge detection and contour extraction are used to identify the body parts in the virtual product image, including arms, legs, torso, etc., and posture estimation algorithms (such as OpenPose, DeepLabCut, etc.) are used to determine the position and posture of the body. When performing expression feature recognition, facial detection algorithms (such as Haar features, HOG+SVM, etc.) are used to locate the facial area, and feature points (such as corners of the eyes, corners of the mouth, etc.) are extracted in the facial area. These feature points can reflect the shape and changes of the face, and expression recognition is performed based on the position and changes of the feature points, and machine learning or deep learning algorithms (such as SVM, convolutional neural network, etc.) are used to recognize expressions.

[0035] As a preferred embodiment of the present invention, the limb feature recognition is performed, edge detection and contour extraction are used to identify the limb parts in the virtual product image, and a posture estimation algorithm is used to determine the position and posture of the limb, which specifically includes the following sub-steps: Downsampling the virtual product image to generate virtual product images of different scales; The edge detection algorithm is used to extract edge information of virtual product images of different scales to obtain multi-scale edge information; The multi-scale edge information is weighted averaged for fusion, and the fusion result is nonlinearly smoothed to obtain an edge image; Performing contour detection on the edge image to obtain a contour set, and then matching the contour set with the limb template to obtain the limb contour; Perform key point detection on the limb contour to generate a key point candidate set; According to the relative position between the human limb proportion and the key point candidate set, the key point candidate set is screened to obtain the limb key points; A limb skeleton is constructed based on the limb key points, and the angle and position relationship of the limbs are calculated according to the limb skeleton for posture estimation.

[0036] In the embodiment of the present invention, multi-scale analysis is utilized to simultaneously capture local details and global structures in virtual product images. By performing edge fusion on data of different resolutions, the system can more accurately extract the edges of limbs, especially the contours of limbs in complex backgrounds, such as interference from clothing and material textures.

[0037] As a preferred embodiment of the present invention, the facial expression feature recognition is performed, the facial region is located using a facial detection algorithm, feature points are extracted from the facial region, and the facial expression recognition is performed based on the position and change of the feature points, specifically comprising the following sub-steps: Use facial detection algorithms to locate facial regions; The facial area is detected using a facial feature point detection algorithm to obtain facial feature key points; Perform local texture verification and geometric consistency correction operations on each facial feature key point to obtain the optimized facial feature key point; According to human facial features, a geometric relationship model is constructed using the distance and angle between optimized facial feature key points; The geometric relationship and dynamic changes of the optimized facial feature key points in the geometric relationship model are combined to generate a high-dimensional target expression feature vector; Using the pre-trained support vector machine as the prediction model, the high-dimensional target expression feature vector is input into the prediction model to obtain the expression recognition result.

[0038] In the embodiment of the present invention, the geometric structure of the face is fully described by the distance and angle between the feature points. This relationship modeling is not only applicable to static images, but can also be extended to dynamic expression analysis, and the analysis of the time dimension is added, such as the displacement, velocity and acceleration of the feature points. These dynamic features can capture the change process of expression from one state to another, making the classification more accurate.

[0039] like Figure 5 As shown, as a preferred embodiment of the present invention, the step of generating a plurality of derivative virtual product images based on body features and expression features specifically includes: S401, retrieve the body movements and facial expressions corresponding to each interaction item; S402, fusing the recognized body features and facial features with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction item; S403: Rendering and adjusting details of the derivative virtual product image.

[0040] In an embodiment of the present invention, it is necessary to set a corresponding body movement and facial expression for each interactive item in advance, and then merge the recognized body features and facial expressions with the retrieved body movements and facial expressions to generate a derivative virtual product image that conforms to the corresponding interactive item. Finally, the derivative virtual product image is rendered and adjusted in detail. The generated virtual product image is rendered using image rendering technology (such as 3D rendering, ray tracing, etc.) to improve its realism and visual effect, and the generated derivative virtual product image is adjusted in detail, such as modifying color, brightness, contrast, etc., to make it more in line with actual needs.

[0041] like Figure 6 As shown, an embodiment of the present invention further provides a virtual product digital image processing system, the system comprising: A virtual image compression module 100 is used to sample a virtual product image, and perform image encoding and compression on the sampled virtual product image; A virtual image rendering module 200, for performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; A key feature recognition module 300 is used to recognize key features in a virtual product image, wherein the key features include body features and facial features; A derived virtual image module 400, for generating a plurality of derived virtual product images based on the body features and the expression features, each derived virtual product image corresponding to an interactive item; Icon button interaction module 500, used to generate an icon button for each interactive item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; The voice information binding module 600 is used to receive a voice binding instruction, bind a voice segment to each interactive item, and play the voice synchronously when the corresponding icon button is triggered.

[0042] As a preferred embodiment of the present invention, the virtual image compression module 100 includes: An image signal conversion unit, used for converting a continuous virtual product image signal into discrete data points according to a set sampling rate; An image data quantization unit, used for quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; The image decoding and restoration unit is used to map the quantized discrete levels back to the frequency domain coefficients, convert the frequency domain coefficients back to the spatial domain to obtain a restored image, perform denoising and color space conversion on the restored image, and obtain a processed virtual product image.

[0043] As a preferred embodiment of the present invention, the virtual image rendering module 200 includes: A virtual image enhancement unit, used for sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of virtual product images; A geometric information transformation unit is used to obtain geometric information of a model corresponding to a virtual product image, transform the model from a world coordinate system to an observation coordinate system, project the 3D model in the observation coordinate system onto a 2D plane, and generate an image of perspective projection or parallel projection; The model ray tracing unit is used to define the light source information, emit light from the camera position, enter the scene for tracing, calculate the intersection points of the light and the model in the scene, as well as the reflection and refraction interaction effects of the light on the model surface, and determine the shadow effect.

[0044] As a preferred embodiment of the present invention, the key feature identification module 300 includes: A limb feature recognition unit is used to perform limb feature recognition, identify limb parts in the virtual product image using edge detection and contour extraction, and determine the position and posture of the limb using a posture estimation algorithm; The expression feature recognition unit is used to perform expression feature recognition, locate the facial area using a facial detection algorithm, extract feature points within the facial area, and perform expression recognition based on the position and changes of the feature points.

[0045] As a preferred embodiment of the present invention, the derived virtual image module 400 includes: An information retrieval unit, used to retrieve the body movements and facial expressions corresponding to each interaction item; A derivative image determination unit, used to fuse the recognized body features and expression features with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction items; The derivative image processing unit is used to render and adjust the details of the derivative virtual product image.

[0046] The above only describes in detail the preferred embodiments of the present invention, which is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

[0047] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0048] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0049] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the disclosure in the specification and examples. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A method for processing a virtual product digital image, characterized in that: The method comprises the following steps: Sampling virtual product images, and performing image encoding and compression on the sampled virtual product images; Performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; Identify key features in the virtual product image, wherein the key features include body features and facial features; Generate a number of derivative virtual product images based on body features and expression features, each derivative virtual product image corresponds to an interactive item; Generate an icon button for each interactive item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; Receive voice binding instructions, bind a voice to each interactive item, and play the voice synchronously when the corresponding icon button is triggered; The step of sampling the virtual product image and encoding and compressing the sampled virtual product image specifically includes: Convert the continuous virtual product image signal into discrete data points according to the set sampling rate; Quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; The quantized discrete levels are mapped back to the frequency domain coefficients, the frequency domain coefficients are converted back to the spatial domain to obtain the restored image, the restored image is denoised and the color space is converted to obtain the processed virtual product image.

2. The method for processing virtual product digital images according to claim 1, characterized in that: Mapping the quantized discrete levels back to frequency domain coefficients, converting the frequency domain coefficients back to the spatial domain, and obtaining a restored image specifically includes the following sub-steps: The quantized discrete levels are compressed using discrete cosine transform to generate a frequency domain coefficient matrix; According to the position of each element in the frequency domain coefficient matrix, a corresponding first adaptive weight is adaptively given, and a weighted calculation is performed to obtain a weighted frequency domain coefficient matrix; Perform nonlinear mapping on the absolute value of each element in the weighted frequency domain coefficient matrix to obtain an enhanced frequency domain coefficient matrix; Counting the energy of the frequency components of the frequency domain coefficient matrix in each direction after the enhancement, and determining the enhancement direction according to the energy distribution; According to the position of each frequency component, calculate the projection value of the frequency component in the current direction; Perform Gaussian distribution processing on the projection values ​​to obtain the directional enhancement weight matrix; Applying the directional enhancement weight matrix to the enhanced frequency domain coefficient matrix to obtain a directional enhanced frequency domain coefficient matrix; The frequency domain coefficient matrix of the direction enhancement is subjected to two-dimensional inverse discrete cosine transform to obtain the restored image.

3. The method for processing virtual product digital images according to claim 2, characterized in that: The denoising and color space conversion of the restored image to obtain a processed virtual product image specifically includes the following sub-steps: Perform multi-scale decomposition on the restored image to decompose the image into multiple levels to obtain high-frequency detail features and low-frequency smooth features; Perform gradient calculation on the restored image to obtain the gradient amplitude of the image; Normalize the gradient amplitude of the image to obtain a gradient normalization result; Adaptively generating a second adaptive weight according to the gradient amplitude in the gradient normalization result; Applying multi-scale decomposition to the restored image to obtain multi-scale detail features, wherein the multi-scale detail features include high-frequency detail features and low-frequency smooth features; The noise suppression algorithm is used to denoise the high-frequency detail features, and the fuzzy processing is used to denoise the low-frequency smooth features to obtain multi-scale denoising results; Combining the multi-scale denoising results according to the second adaptive weight to obtain a denoised image; The denoised image is converted into a color space to obtain a processed virtual product image.

4. The method for processing virtual product digital images according to claim 3, characterized in that: The step of performing image enhancement and 3D model rendering on the compressed virtual product image specifically includes: Sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of virtual product images; Obtain the geometric information of the model corresponding to the virtual product image, transform the model from the world coordinate system to the observation coordinate system, project the 3D model in the observation coordinate system onto a 2D plane, and generate a perspective projection or parallel projection image; Define the light source information, emit light from the camera position, enter the scene for tracking, calculate the intersection point of the light and the model in the scene, as well as the reflection and refraction interaction effects of the light on the model surface, and determine the shadow effect.

5. The method for processing virtual product digital images according to claim 4, characterized in that: The step of identifying key features in the virtual product image specifically includes: Perform limb feature recognition, use edge detection and contour extraction to identify limb parts in virtual product images, and use posture estimation algorithms to determine the position and posture of limbs; Perform facial expression feature recognition, use facial detection algorithm to locate the facial area, extract feature points in the facial area, and perform facial expression recognition based on the position and changes of feature points.

6. The method for processing virtual product digital images according to claim 5, characterized in that: The limb feature recognition is performed by using edge detection and contour extraction to identify the limb parts in the virtual product image, and using a posture estimation algorithm to determine the position and posture of the limb, which specifically includes the following sub-steps: Downsampling the virtual product image to generate virtual product images of different scales; The edge detection algorithm is used to extract edge information of virtual product images of different scales to obtain multi-scale edge information; The multi-scale edge information is weighted averaged for fusion, and the fusion result is nonlinearly smoothed to obtain an edge image; Performing contour detection on the edge image to obtain a contour set, and then matching the contour set with the limb template to obtain the limb contour; Perform key point detection on the limb contour to generate a key point candidate set; According to the relative position between the human limb proportion and the key point candidate set, the key point candidate set is screened to obtain the limb key points; A limb skeleton is constructed based on the limb key points, and the angle and position relationship of the limbs are calculated according to the limb skeleton for posture estimation.

7. The method for processing virtual product digital images according to claim 6, characterized in that: The facial expression feature recognition comprises locating the facial area using a facial detection algorithm, extracting feature points in the facial area, and performing facial expression recognition based on the positions and changes of the feature points, which specifically comprises the following sub-steps: Use facial detection algorithms to locate facial regions; The facial area is detected using a facial feature point detection algorithm to obtain facial feature key points; Perform local texture verification and geometric consistency correction operations on each facial feature key point to obtain the optimized facial feature key point; According to human facial features, a geometric relationship model is constructed using the distances and angles between optimized facial feature key points; The geometric relationship and dynamic changes of the optimized facial feature key points in the geometric relationship model are combined to generate a high-dimensional target expression feature vector; Using the pre-trained support vector machine as the prediction model, the high-dimensional target expression feature vector is input into the prediction model to obtain the expression recognition result.

8. The method for processing virtual product digital images according to claim 7, characterized in that: The step of generating a plurality of derivative virtual product images based on the body features and expression features specifically includes: Retrieve the body movements and facial expressions corresponding to each interaction item; The recognized body features and facial features are integrated with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction items; The derived virtual product image is rendered and detail adjusted.

9. A virtual product digital image processing system, characterized in that: The system applies the virtual product digital image processing method according to any one of claims 1 to 8, and the system comprises: A virtual image compression module, used for sampling virtual product images, and performing image encoding and compression on the sampled virtual product images; A virtual image rendering module, used for performing image enhancement and 3D model rendering on the compressed virtual product image, wherein the 3D model rendering includes geometric calculation and ray tracing; A key feature recognition module, used to recognize key features in a virtual product image, wherein the key features include body features and facial features; A derived virtual image module, used to generate a number of derived virtual product images based on body features and expression features, each derived virtual product image corresponding to an interactive entry; An icon button interaction module is used to generate an icon button for each interaction item, and the user clicks the icon button to achieve dynamic interaction with the virtual product image; The voice information binding module is used to receive voice binding instructions, bind a voice segment to each interactive item, and play the voice synchronously when the corresponding icon button is triggered.

10. The virtual product digital image processing system according to claim 9, characterized in that: The virtual image compression module comprises: An image signal conversion unit, used for converting a continuous virtual product image signal into discrete data points according to a set sampling rate; An image data quantization unit, used for quantizing the converted data into discrete grayscale values ​​or color values, and encoding and compressing the quantized data into digital signals; An image decoding and restoration unit is used to map the quantized discrete levels back to the frequency domain coefficients, convert the frequency domain coefficients back to the spatial domain to obtain a restored image, perform denoising and color space conversion on the restored image, and obtain a processed virtual product image; The virtual image rendering module comprises: A virtual image enhancement unit, used for sharpening, contrast adjustment, color correction, color saturation adjustment, and spatial domain enhancement of virtual product images; A geometric information transformation unit is used to obtain geometric information of a model corresponding to a virtual product image, transform the model from a world coordinate system to an observation coordinate system, project the 3D model in the observation coordinate system onto a 2D plane, and generate an image of perspective projection or parallel projection; Model ray tracing unit, used to define light source information, emit light from the camera position, enter the scene for tracing, calculate the intersection point of light with the model in the scene, and the reflection and refraction interaction effects of light on the model surface, and determine the shadow effect; The key feature recognition module includes: A limb feature recognition unit is used to perform limb feature recognition, identify limb parts in the virtual product image using edge detection and contour extraction, and determine the position and posture of the limb using a posture estimation algorithm; The expression feature recognition unit is used to perform expression feature recognition, locate the facial area using a facial detection algorithm, extract feature points in the facial area, and perform expression recognition based on the position and change of the feature points; The derived virtual image module comprises: An information retrieval unit, used to retrieve the body movements and facial expressions corresponding to each interaction item; A derivative image determination unit, used to fuse the recognized body features and expression features with the retrieved body movements and facial expressions to generate a derivative virtual product image that meets the corresponding interaction items; The derivative image processing unit is used to render and adjust the details of the derivative virtual product image.

Images