Metacosm virtual digital human manufacturing method and system

Abstract

The invention provides a meta-universe virtual digital human making method and system, and relates to the technical field of human modeling. The method comprises the following steps: collecting a body feature data set of a user at multiple angles and under different illumination conditions by using sensing equipment; integrating and optimizing the body feature data set to obtain a group of final body features; determining an initial head shape and an initial body shape of the character model according to the final body characteristics; correcting the initial head shape and the initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape; fitting the corrected head form and the corrected body form to obtain an initial figure model; performing data feature extraction and segmentation on a target person, and extracting key features; extracting figure details according to the key features; correcting the initial character model according to the character details to obtain an intermediate character model; and obtaining a virtual digital human according to the intermediary character model. The problem of low modeling accuracy in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of character modeling, and in particular to a method and system for producing a metaverse virtual digital human. Background Art

[0002] The metaverse is essentially a virtualization and digitization process of the real world. The metaverse needs to include various forms of virtual digital people. Virtual digital people refer to virtual characters that exist in the physical world and have a digital appearance. The existing methods for making virtual digital people mainly include face pinching modeling. When making virtual digital people by artificial face pinching modeling, the production cycle is long, and it is difficult to quickly obtain a virtual digital person with unique features and expected appearance. For the above-mentioned problems, the invention patent with patent publication number CN 117392330 B discloses a method and system for making virtual digital people in the metaverse. The current patent uses facial scan images to determine the head shape, determines the body shape of the person according to height, weight and gender to determine the preliminary character model, and automatically pinches the face of the preliminary character model and accepts the facial features to obtain a virtual digital person. In the above content, the processing and feature recognition of facial scan images may be affected by factors such as ambient light and changes in facial expressions, resulting in insufficient recognition accuracy and subjective and inaccurate information such as gender, height, and weight input by the user, which affects the realism of the final generated character model. Summary of the invention

[0003] In order to overcome the deficiencies of the prior art, the purpose of the present invention is to provide a method and system for producing a metaverse virtual digital human.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A method for producing a metaverse virtual digital human, comprising:

[0006] Using a sensor device to collect a user's body feature data set at multiple angles and under different lighting conditions, the body feature data includes: facial data, body shape information, and skin color features;

[0007] Integrating and optimizing the body feature data set to obtain a final set of body features;

[0008] Determining an initial head shape and an initial body shape of the character model according to the final body features;

[0009] Obtain the target person's physiological data based on biometric technology;

[0010] Correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape;

[0011] Fitting the corrected head shape and the corrected body shape to obtain an initial character model;

[0012] Extracting and segmenting data features of the target person to extract key features;

[0013] Extracting character details according to the key features, the character details including: facial details, clothing details, jewelry details and emotional details;

[0014] Modifying the initial character model according to the character details to obtain an intermediate character model;

[0015] The facial details, clothing details, jewelry details and emotional details of the intermediate character model are judged for rationality in a linked manner to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human.

[0016] Preferably, the body feature data set is integrated and optimized to obtain a final set of body features, including:

[0017] Normalizing the data in the body feature data set to obtain normalized data;

[0018] Performing noise reduction processing on the normalized data to obtain noise-reduced data;

[0019] Based on the CNN neural network, the denoised data is subjected to multiple image feature extractions to obtain an image feature set, wherein the image feature set includes: a facial feature set, a body feature set and a skin color feature set;

[0020] Performing similarity calculation of feature values ​​under different lighting conditions on various feature sets in the image feature set to obtain a first similarity feature set;

[0021] Using a weighted average method to fuse various similarity feature sets in the first similarity feature set to obtain a fused feature set, wherein the fused feature set includes a facial fusion feature, a body fusion feature, and a skin color fusion feature;

[0022] Calculating the similarities between various fused features in the fused feature set to obtain a second similarity set;

[0023] The rationality between the second similarity sets is judged to obtain a judgment result. If the judgment result is reasonable, the final body feature is obtained. If the judgment result is unreasonable, return to the step of "based on the CNN neural network, perform multiple image feature extraction on the denoised data to obtain an image feature set".

[0024] Preferably, the expression of the normalized data is:

[0025]

[0026] Among them, X ′ is the normalized data, X is the body feature data set, μ is the mean of the body feature data set, and σ is the standard deviation of the body feature data set.

[0027] Preferably, the step of obtaining the target person's physiological data based on biometric recognition technology includes:

[0028] 3D imaging data of a target person using a 3D scanner, wherein the 3D imaging data includes: 3D facial data and 3D torso data;

[0029] Extracting geometric features from the 3D surface data to obtain head shape contour feature data, wherein the head shape contour data includes: skull height, skull width, and forehead inclination;

[0030] Extracting proportion features from the 3D torso data to obtain torso feature data, wherein the torso feature data includes: arm spread length, chest-to-waist ratio, and leg length-to-height ratio;

[0031] The torso feature data and the head shape contour feature data are integrated to obtain physiological data.

[0032] Preferably, the correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape comprises:

[0033] Determine a mapping relationship between the head shape parameters of the initial head shape and the head shape contour feature data to obtain a first mapping relationship table;

[0034] Calculating the difference between the head shape contour feature data and the initial head shape to obtain a head shape correction vector;

[0035] Correcting the initial head shape according to the head shape correction vector and the first mapping relationship table to obtain a corrected head shape;

[0036] Determine a mapping relationship between the body shape parameters of the initial body shape and the trunk feature data to obtain a second mapping relationship table;

[0037] Calculating the difference between the trunk feature data and the initial body shape to obtain a body shape correction vector;

[0038] The initial body shape is corrected according to the body shape correction vector and the second mapping relationship table to obtain a corrected body shape.

[0039] Preferably, the facial details, clothing details, jewelry details and emotional details of the intermediate character model are judged for rationality in a linked manner to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If the judgment result set contains unreasonable details, the corresponding details are corrected to obtain a virtual digital human, including:

[0040] Extracting detail feature vectors of the facial details, clothing details, jewelry details and emotional details and forming a multi-dimensional detail feature vector;

[0041] Annotating each detail vector in the multidimensional detail feature vector with a label;

[0042] Perform multiple linkage rationality judgments based on the annotated detailed feature vectors to obtain multiple corresponding linkage scores;

[0043] According to the correlation of the detailed feature vectors of the annotated labels, the corresponding scoring threshold sets are set in different combinations;

[0044] The multiple corresponding linkage scores are compared with the score threshold set. If the comparison results are all reasonable, a virtual digital human is obtained. If the comparison results are unreasonable, the corresponding detail features are confirmed and the details are corrected.

[0045] A metaverse virtual digital human production system, comprising:

[0046] A first data acquisition module is used to collect a user's body feature data set using a sensor device at multiple angles and under different lighting conditions, wherein the body feature data includes: facial data, body shape information, and skin color features;

[0047] An integration module, used for integrating and optimizing the body feature data set to obtain a set of final body features;

[0048] An initial head shape and body shape determining module, used to determine the initial head shape and initial body shape of the character model according to the final body features;

[0049] A second data acquisition module is used to acquire physiological data of the target person based on biometric recognition technology;

[0050] A correction module, used for correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape;

[0051] An initial character model determination module, used for fitting the modified head shape and the modified body shape to obtain an initial character model;

[0052] A segmentation module is used to extract data features of the target person and perform segmentation to extract key features;

[0053] A detail determination module, used to extract character details according to the key features, wherein the character details include: facial details, clothing details, jewelry details and emotional details;

[0054] An intermediate model determination module, used for correcting the initial character model according to the character details to obtain an intermediate character model;

[0055] The virtual digital human determination module is used to perform linkage rationality judgment on the facial details, clothing details, jewelry details and emotional details of the intermediate character model to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human.

[0056] The present invention discloses the following technical effects:

[0057] The present invention provides a method and system for producing a metaverse virtual digital human, the method comprising: using a sensor device to collect a user's body feature data set at multiple angles and under different lighting conditions, the body feature data comprising: facial data, body shape information and skin color features; integrating and optimizing the body feature data set to obtain a set of final body features; determining an initial head shape and an initial body shape of a character model according to the final body features; obtaining physiological data of a target person according to a biometric recognition technology; correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape; fitting the corrected head shape and initial body shape The corrected body shape is used to obtain an initial character model; data features are extracted and segmented for the target character to extract key features; character details are extracted based on the key features, and the character details include: facial details, clothing details, jewelry details, and emotional details; the initial character model is corrected based on the character details to obtain an intermediate character model; the facial details, clothing details, jewelry details, and emotional details of the intermediate character model are linked for rationality judgment to obtain a judgment result set, and if the judgment result set is all reasonable, a virtual digital person is obtained; if there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital person. The present invention ensures that the user's physical feature data (such as face, body shape, and skin color) is reflected with high precision by collecting physical feature data from multiple angles and under different lighting conditions. This method reduces model deviations caused by light changes and different angles, thereby enhancing the realism of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0059] Figure 1 A flow chart of a method for producing a metaverse virtual digital human provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] 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.

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] like Figure 1 As shown, the present invention provides a method for producing a virtual digital human in a metaverse, comprising:

[0063] Step 100: Using a sensor device to collect a user's body feature data set at multiple angles and under different lighting conditions, the body feature data including: facial data, body shape information, and skin color features;

[0064] Specifically, multiple sensing devices (such as depth cameras, high-resolution cameras, RGB-D sensors, etc.) are used to collect user's physical feature data.

[0065] The sensing device should have an automatic mode to automatically adapt to environmental changes and ensure accurate capture of the following data under different lighting conditions and angles:

[0066] Facial feature data: including eyes, nose, lips, chin, etc. Each feature should be collected at multiple camera angles to generate a complete 3D model.

[0067] Posture information: Use motion capture technology (such as IMU-based sensors or optical tracking) to collect the user's posture, including standing, sitting, and dynamic movements.

[0068] Skin color characteristics: Use a color difference sensor to obtain skin color information and describe it according to international color standards (such as RGB or CIELab).

[0069] Step 200: Integrate and optimize the body feature data set to obtain a final set of body features;

[0070] Step 300: determining an initial head shape and an initial body shape of the character model according to the final body features;

[0071] Step 400: Acquire physiological data of the target person according to biometric recognition technology;

[0072] Step 500: Correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape;

[0073] Step 600: fitting the corrected head shape and the corrected body shape to obtain an initial character model;

[0074] Step 700: extracting data features of the target person and performing segmentation to extract key features;

[0075] Step 800: extracting character details according to the key features, the character details including: facial details, clothing details, jewelry details and emotional details;

[0076] Step 900: modifying the initial character model according to the character details to obtain an intermediate character model;

[0077] Step 100: Perform linkage rationality judgment on the facial details, clothing details, jewelry details and emotional details of the intermediate character model to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital person is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital person.

[0078] Furthermore, the body feature data set is integrated and optimized to obtain a final set of body features, including:

[0079] Normalizing the data in the body feature data set to obtain normalized data;

[0080] Specifically, the normalization method is: such as Z-score normalization or Min-Max normalization.

[0081] Performing noise reduction processing on the normalized data to obtain noise-reduced data;

[0082] Based on the CNN neural network, the denoised data is subjected to multiple image feature extractions to obtain an image feature set, wherein the image feature set includes: a facial feature set, a body feature set and a skin color feature set;

[0083] Performing similarity calculation of feature values ​​under different lighting conditions on various feature sets in the image feature set to obtain a first similarity feature set;

[0084] Specifically, the similarity of the feature values ​​under different lighting conditions is calculated to determine their consistency under changes in the external environment.

[0085]

[0086] Among them, S(i,j) is the similarity between feature Fi and feature Fj, Fi is feature i, Fj is feature j, and n is the feature dimension.

[0087] Using a weighted average method to fuse various similarity feature sets in the first similarity feature set to obtain a fused feature set, wherein the fused feature set includes a facial fusion feature, a body fusion feature, and a skin color fusion feature;

[0088] Calculating the similarities between various fused features in the fused feature set to obtain a second similarity set;

[0089] The rationality between the second similarity sets is judged to obtain a judgment result. If the judgment result is reasonable, the final body feature is obtained. If the judgment result is unreasonable, return to the step of "based on the CNN neural network, perform multiple image feature extraction on the denoised data to obtain an image feature set".

[0090] Specifically, the expression of the normalized data is:

[0091]

[0092] Among them, X ′ is the normalized data, X is the body feature data set, μ is the mean of the body feature data set, and σ is the standard deviation of the body feature data set.

[0093] Furthermore, the step of obtaining the target person's physiological data based on biometric technology includes:

[0094] 3D imaging data of a target person using a 3D scanner, wherein the 3D imaging data includes: 3D facial data and 3D torso data;

[0095] Extracting geometric features from the 3D surface data to obtain head shape contour feature data, wherein the head shape contour data includes: skull height, skull width, and forehead inclination;

[0096] Extracting proportion features from the 3D torso data to obtain torso feature data, wherein the torso feature data includes: arm spread length, chest-to-waist ratio, and leg length-to-height ratio;

[0097] The torso feature data and the head shape contour feature data are integrated to obtain physiological data.

[0098] Specifically, skull height: extract key points from the 3D facial point cloud, including the coordinates of the top (highest point) and the mandible (lowest point).

[0099] Skull width: Skull width refers to the horizontal distance between the two sides of the head (behind the ears).

[0100] Forehead tilt: Forehead tilt is the angle between the forehead and the vertical line, which can reflect whether the head is tilted forward or backward.

[0101] Arm's spread length: Arm's spread length is the horizontal distance from one fingertip to the other.

[0102] The chest-to-waist ratio refers to the ratio of chest circumference to waist circumference, which is used to reflect body shape characteristics.

[0103] The leg length to height ratio refers to the ratio of leg length to total height, which can reflect the individual's body characteristics.

[0104] Furthermore, the step of correcting the initial head shape and the initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape includes:

[0105] Determine a mapping relationship between the head shape parameters of the initial head shape and the head shape contour feature data to obtain a first mapping relationship table;

[0106] Specifically, by determining the mapping relationship table between the initial head shape and the head shape contour feature data (the first mapping relationship table) and the mapping relationship table between the initial body shape and the trunk feature data (the second mapping relationship table), wide adaptability can be ensured to a certain extent. Such mapping can make different features correspond reasonably and form an effective adjustment basis.

[0107] Calculating the difference between the head shape contour feature data and the initial head shape to obtain a head shape correction vector;

[0108] Correcting the initial head shape according to the head shape correction vector and the first mapping relationship table to obtain a corrected head shape;

[0109] Specifically, the input data are: initial head shape parameters: including relevant head shape feature parameters, such as head circumference, skull height, skull width, forehead inclination, etc. Head shape contour feature data: feature parameters of the target head shape obtained from 3D scanning or other methods. Head shape correction vector: based on the difference between the head shape contour feature and the initial head shape, indicating the adjustment that needs to be made. The first mapping relationship table: provides the relationship between the initial head shape parameters and the target head shape features.

[0110] The obtained head shape correction vector V should include the differences of each feature on the X, Y, and Z axes. Usually, these differences are abstracted into a set of values ​​for the next correction.

[0111] Each initial feature parameter is combined with the mapping relationship, and the initial parameters are updated to obtain the corrected head shape.

[0112] In this process, linear weighting or the use of a smooth transformation function can be considered to ensure the smoothness and naturalness of the correction result.

[0113] Assuming that the initial head shape parameters include height (250mm), width (180mm), and depth (160mm), the head shape correction vector is: [3mm, -2mm, 5mm]. If the first mapping relationship table is linear (1.0 times), then: new head shape height: 250+3×1=253mm, new head shape width: 180-2×1=178mm, new head shape depth: 160+5×1=165mm.

[0114] The final generated corrected head shape parameters are: [253mm, 178mm, 165mm].

[0115] Determine a mapping relationship between the body shape parameters of the initial body shape and the trunk feature data to obtain a second mapping relationship table;

[0116] Calculating the difference between the trunk feature data and the initial body shape to obtain a body shape correction vector;

[0117] The initial body shape is corrected according to the body shape correction vector and the second mapping relationship table to obtain a corrected body shape.

[0118] Furthermore, the facial details, clothing details, jewelry details and emotional details of the intermediate character model are linked to make a rationality judgment to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human, including:

[0119] Extracting detail feature vectors of the facial details, clothing details, jewelry details and emotional details and forming a multi-dimensional detail feature vector;

[0120] Annotating each detail vector in the multidimensional detail feature vector with a label;

[0121] Specifically, develop a user interface or use automated tools to allow experts to annotate each detailed feature vector with labels. For example, classify and label each facial feature, clothing type, jewelry style, and emotional state (such as "natural", "exaggerated", "suitable for the occasion", etc.).

[0122] Use a judgment mechanism (such as a rule engine or deep learning model) to judge the rationality of detailed features. Measure the logical consistency and relative adaptability between each detail.

[0123] Compare the relationships between features, for example, matching emotions to facial expressions, matching clothing styles to occasions, etc.

[0124] Perform multiple linkage rationality judgments based on the annotated detailed feature vectors to obtain multiple corresponding linkage scores;

[0125] According to the correlation of the detailed feature vectors of the annotated labels, the corresponding scoring threshold sets are set in different combinations;

[0126] The multiple corresponding linkage scores are compared with the score threshold set. If the comparison results are all reasonable, a virtual digital human is obtained. If the comparison results are unreasonable, the corresponding detail features are confirmed and the details are corrected.

[0127] Specifically, the above combination does not necessarily have no combination of two details. There may be a combination of clothing and emotions, or a combination of clothing, emotions and jewelry, etc.

[0128] This embodiment also provides a metaverse virtual digital human production system, including:

[0129] A first data acquisition module is used to collect a user's body feature data set using a sensor device at multiple angles and under different lighting conditions, wherein the body feature data includes: facial data, body shape information, and skin color features;

[0130] An integration module, used for integrating and optimizing the body feature data set to obtain a set of final body features;

[0131] An initial head shape and body shape determining module, used to determine the initial head shape and initial body shape of the character model according to the final body features;

[0132] A second data acquisition module is used to acquire physiological data of the target person based on biometric recognition technology;

[0133] A correction module, used for correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape;

[0134] An initial character model determination module, used for fitting the modified head shape and the modified body shape to obtain an initial character model;

[0135] A segmentation module is used to extract data features of the target person and perform segmentation to extract key features;

[0136] A detail determination module, used to extract character details according to the key features, wherein the character details include: facial details, clothing details, jewelry details and emotional details;

[0137] An intermediate model determination module, used for correcting the initial character model according to the character details to obtain an intermediate character model;

[0138] The virtual digital human determination module is used to perform linkage rationality judgment on the facial details, clothing details, jewelry details and emotional details of the intermediate character model to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human.

[0139] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0140] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for producing a virtual digital human in a metaverse, characterized in that: include: Using a sensor device to collect a user's body feature data set at multiple angles and under different lighting conditions, the body feature data includes: facial data, body shape information, and skin color features; Integrating and optimizing the body feature data set to obtain a final set of body features; Determining an initial head shape and an initial body shape of the character model according to the final body features; Obtain the target person's physiological data based on biometric technology; Correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape; Fitting the corrected head shape and the corrected body shape to obtain an initial character model; Extracting and segmenting data features of the target person to extract key features; Extracting character details according to the key features, the character details including: facial details, clothing details, jewelry details and emotional details; Modifying the initial character model according to the character details to obtain an intermediate character model; The facial details, clothing details, jewelry details and emotional details of the intermediate character model are judged for rationality in a linked manner to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human.

2. A method for producing a virtual digital human in a metaverse according to claim 1, characterized in that: The body feature data set is integrated and optimized to obtain a set of final body features, including: Normalizing the data in the body feature data set to obtain normalized data; Performing noise reduction processing on the normalized data to obtain noise-reduced data; Based on the CNN neural network, the denoised data is subjected to multiple image feature extractions to obtain an image feature set, wherein the image feature set includes: a facial feature set, a body feature set and a skin color feature set; Performing similarity calculation of feature values ​​under different lighting conditions on various feature sets in the image feature set to obtain a first similarity feature set; Using a weighted average method to fuse various similarity feature sets in the first similarity feature set to obtain a fused feature set, wherein the fused feature set includes a facial fusion feature, a body fusion feature, and a skin color fusion feature; Calculating the similarities between various fused features in the fused feature set to obtain a second similarity set; The rationality between the second similarity sets is judged to obtain a judgment result. If the judgment result is reasonable, the final body feature is obtained. If the judgment result is unreasonable, return to the step of "based on the CNN neural network, perform multiple image feature extraction on the denoised data to obtain an image feature set".

3. A method for producing a virtual digital human in a metaverse according to claim 2, characterized in that: The expression of the normalized data is: Among them, X ′ is the normalized data, X is the body feature data set, μ is the mean of the body feature data set, and σ is the standard deviation of the body feature data set.

4. The method for producing a virtual digital human in a metaverse according to claim 1, characterized in that: The method of obtaining the target person's physiological data based on biometric technology includes: 3D imaging data of a target person using a 3D scanner, wherein the 3D imaging data includes: 3D facial data and 3D torso data; Extracting geometric features from the 3D surface data to obtain head shape contour feature data, wherein the head shape contour data includes: skull height, skull width, and forehead inclination; Extracting proportion features from the 3D torso data to obtain torso feature data, wherein the torso feature data includes: arm spread length, chest-to-waist ratio, and leg length-to-height ratio; The torso feature data and the head shape contour feature data are integrated to obtain physiological data.

5. A method for producing a virtual digital human in a metaverse according to claim 4, characterized in that: The step of correcting the initial head shape and the initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape comprises: Determine a mapping relationship between the head shape parameters of the initial head shape and the head shape contour feature data to obtain a first mapping relationship table; Calculating the difference between the head shape contour feature data and the initial head shape to obtain a head shape correction vector; Correcting the initial head shape according to the head shape correction vector and the first mapping relationship table to obtain a corrected head shape; Determine a mapping relationship between the body shape parameters of the initial body shape and the trunk feature data to obtain a second mapping relationship table; Calculating the difference between the trunk feature data and the initial body shape to obtain a body shape correction vector; The initial body shape is corrected according to the body shape correction vector and the second mapping relationship table to obtain a corrected body shape.

6. The method for producing a virtual digital human in a metaverse according to claim 1, characterized in that: The linkage rationality judgment of the facial details, clothing details, jewelry details and emotional details of the intermediate character model is performed to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital person is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital person, including: Extracting detail feature vectors of the facial details, clothing details, jewelry details and emotional details and forming a multi-dimensional detail feature vector; Annotating each detail vector in the multidimensional detail feature vector with a label; Perform multiple linkage rationality judgments based on the annotated detailed feature vectors to obtain multiple corresponding linkage scores; According to the correlation of the detailed feature vectors of the annotated labels, the corresponding scoring threshold sets are set in different combinations; The multiple corresponding linkage scores are compared with the score threshold set. If the comparison results are all reasonable, a virtual digital human is obtained. If the comparison results are unreasonable, the corresponding detail features are confirmed and the details are corrected.

7. A metaverse virtual digital human production system, characterized in that: include: A first data acquisition module is used to collect a user's body feature data set using a sensor device at multiple angles and under different lighting conditions, wherein the body feature data includes: facial data, body shape information, and skin color features; An integration module, used for integrating and optimizing the body feature data set to obtain a set of final body features; An initial head shape and body shape determining module, used to determine the initial head shape and initial body shape of the character model according to the final body features; A second data acquisition module is used to acquire physiological data of the target person based on biometric recognition technology; A correction module, used for correcting the initial head shape and initial body shape according to the physiological data to obtain a corrected head shape and a corrected body shape; An initial character model determination module, used for fitting the modified head shape and the modified body shape to obtain an initial character model; A segmentation module is used to extract data features of the target person and perform segmentation to extract key features; A detail determination module, used to extract character details according to the key features, wherein the character details include: facial details, clothing details, jewelry details and emotional details; An intermediate model determination module, used for correcting the initial character model according to the character details to obtain an intermediate character model; The virtual digital human determination module is used to perform linkage rationality judgment on the facial details, clothing details, jewelry details and emotional details of the intermediate character model to obtain a judgment result set. If the judgment result set is all reasonable, a virtual digital human is obtained. If there are unreasonable details in the judgment result set, the corresponding details are corrected to obtain a virtual digital human.

Images