An AI digital human modeling method and system based on user portraits

Abstract

The present invention relates to the technical field of digital human modeling, and specifically discloses an AI digital human modeling method and system based on user portraits, which are used to solve the problem that existing digital humans are difficult to match the current appearance characteristics of users and the system cannot accurately identify users; the present invention obtains the first facial features and the first digital feature parameters of the user, constructs a first-level differential feature analysis model to obtain the first differential feature parameters of the user, and based on the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, obtains the second differential feature parameters of the user, constructs a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, performs user facial change recognition and matching, and marks and warns users with difficult matching, realizing the accurate identification and dynamic tracking of the facial health and aging characteristics of users.

Description

Technical Field

[0001] The present invention relates to the technical field of digital human modeling, and more specifically, to an AI digital human modeling method and system based on user portraits. Background Art

[0002] In the long-term medical and health management scenario, AI digital human modeling based on user portraits is widely used in facial recognition, pathological analysis, and personalized health consultations. When a user initially establishes a health record, a three-dimensional digital human of the user is generated through the uploaded two-dimensional facial image for identifying user characteristics and tracking health changes. However, as the user ages, their facial features (such as loose skin, deepening wrinkles, and changes in fat distribution) gradually change significantly, while the generated digital human model is based on the original data and lacks the ability to dynamically model the natural aging process. For example, a digital human generated from a facial image uploaded by a user at the age of 30 is used to manage facial health. However, as the user ages to 35, their skin becomes more saggy, eye bags become obvious, and nasolabial folds deepen, but the system still relies on the digital human model generated from the initial image five years ago and fails to update dynamically. This difference makes it difficult for the digital human to match the user's current appearance characteristics, and the system cannot accurately identify the user. To solve the above problems, a technical solution is provided now. Summary of the Invention

[0003] To overcome the above-mentioned defects of the prior art, the present invention provides an AI digital human modeling method and system based on user portraits to solve the problem that existing digital humans are difficult to match the user's current appearance characteristics and the system cannot accurately identify the user, so as to solve the problems raised in the above background art.

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

[0005] An AI digital human modeling method based on user portraits, comprising the following steps:

[0006] Step 1, obtain the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, perform identification processing on the original two-dimensional data information to obtain the first facial feature parameters of the user, and obtain the first digital feature parameters based on the original digital human facial information;

[0007] Step 2, construct a first-level discrimination feature analysis model based on the first facial features and the first digital feature parameters to obtain the first discrimination feature parameters of the user; the first discrimination feature parameters include the first facial aging feature parameters and the first health feature parameters; the formula of the first-level discrimination feature analysis model is:

[0008] ;

[0009] ;

[0010] Wherein: is the first facial aging characteristic parameter, is the first health characteristic parameter, is the distance between the inner corners of the user's eyes, is the distance between the left inner corner of the user's eye and the tip of the nose, is the distance between the right inner corner of the user's eye and the tip of the nose, is the distance between the left corner of the user's mouth and the tip of the nose, is the distance between the right corner of the user's mouth and the tip of the nose, is the pixel value of the i-th point in the original two-dimensional facial data information, is the first distribution coefficient, is the first characteristic mean coefficient, is the number of pixel points in the original two-dimensional facial data information;

[0011] Step 3: Obtain the user's first two-dimensional facial data information and the first digital human facial information generated based on the first two-dimensional facial data information, process the first two-dimensional facial data information and the first digital human facial information respectively, and obtain the user's second discrimination characteristic parameter;

[0012] Step 4: Construct a facial feature change matching model according to the first discrimination characteristic parameter and the second discrimination characteristic parameter to obtain the user's facial matching index, perform user facial change recognition and matching, and mark and warn users with difficult matching.

[0013] As a further solution of the present invention, in step 1, obtain the user's original two-dimensional facial data information and the original digital human information generated based on the two-dimensional facial data information, perform recognition processing on the original two-dimensional data information to obtain the user's first facial characteristic parameter, and the first facial characteristic parameter includes the pixel value of the original two-dimensional facial data information, the first distribution coefficient and the first characteristic mean coefficient: obtain the user's frontal two-dimensional facial image in the original two-dimensional facial data information, perform segmentation processing on the user's frontal two-dimensional facial image, segment the frontal two-dimensional facial image into several facial feature regions, select several feature points in each facial feature region, calculate the variance of the pixel values of all feature points to obtain the first distribution coefficient of each facial feature region, and calculate the average value by summing the pixel values of all feature points to obtain the first characteristic mean coefficient.

[0014] As a further solution of the present invention, in step 1, the first digital feature parameters are obtained based on the original digital human facial information: the original digital human facial information is obtained, and the three-dimensional coordinates of the key points of the original digital human facial information are located, that is, the three-dimensional coordinates of the user's eye corners, the tip of the nose, and the corners of the mouth, and the distances between the two eye corners of the user, the distances between the left and right eye corners and the tip of the nose, and the distances between the left and right corners of the mouth and the tip of the nose are respectively calculated and obtained as the first digital feature parameters.

[0015] As a further solution of the present invention, in step 3, the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information are obtained, and the first two-dimensional facial data information and the first digital human facial information are respectively processed to obtain the second difference feature parameters of the user: the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information are obtained, the first two-dimensional facial data information is identified and processed to obtain the second facial feature parameters of the user, and the second digital feature parameters are obtained based on the first digital human facial information; a secondary difference feature analysis model is constructed based on the second facial feature and the second digital feature parameters to obtain the second difference feature parameters of the user; the second difference feature parameters include the second facial aging feature parameters and the second health feature parameters; the formula of the secondary difference feature analysis model is:

[0016] ;

[0017] ;

[0018] In the formula: is the second facial aging feature parameter, is the second health feature parameter, is the second distance between the two eye corners of the user, is the second distance between the left eye corner of the user and the tip of the nose, is the second distance between the right eye corner of the user and the tip of the nose, is the second distance between the left corner of the user's mouth and the tip of the nose, is the pixel value of the i-th point in the first two-dimensional facial data information, is the second distribution coefficient, is the second feature mean coefficient, is the number of pixel points in the first two-dimensional facial data information.

[0019] As a further solution of the present invention, in step 4, a facial feature change matching model is constructed based on the first difference feature parameter and the second difference feature parameter to obtain a user facial matching index, and user facial change recognition and matching are performed, and users with difficult matching are marked and warned: Obtain the first facial aging feature parameter and the first health feature parameter in the first difference feature parameter, the second facial aging feature parameter and the second health feature parameter in the second difference feature parameter, and construct a facial feature change matching model based on the first difference feature parameter and the second difference feature parameter to obtain a user facial matching index. The formula of the facial feature change matching model is:

[0020] ;

[0021] In the formula: is the user facial matching index, is the first facial aging feature parameter, is the first health feature parameter, is the second facial aging feature parameter, is the second health feature parameter;

[0022] Obtain the user facial matching index, compare the user facial matching index with a preset facial matching threshold. If the user facial matching index is greater than or equal to the preset facial matching threshold, mark the user facial image as difficult to recognize; if the user facial matching index is less than the preset facial matching threshold, there is no need to mark the user facial image as difficult to recognize.

[0023] An AI digital human modeling system based on user portraits, which is used to implement the above-mentioned AI digital human modeling method based on user portraits, includes a processor and a data acquisition and feature extraction module, a first differential feature parameter analysis module, a second differential feature parameter analysis module, a facial feature change matching evaluation module, and a matching warning module that are communicatively connected to the processor; the data acquisition and feature extraction module is used to obtain the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, perform identification processing on the original two-dimensional data information to obtain the first facial feature parameters of the user, and obtain the first digital feature parameters based on the original digital human facial information; the first differential feature parameter analysis module constructs a first-level differential feature analysis model based on the first facial feature and the first digital feature parameters to obtain the first differential feature parameters of the user; the first differential feature parameter analysis module is used to obtain the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, process the first two-dimensional facial data information and the first digital human facial information respectively to obtain the second differential feature parameters of the user; the facial feature change matching evaluation module is used to construct a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, and perform user facial change recognition and matching; the matching warning module is used to mark and warn users with difficult matching.

[0024] The technical effects and advantages of the AI digital human modeling method and system based on user portraits of the present invention: By obtaining the first facial feature and the first digital feature parameters of the user, constructing a first-level differential feature analysis model to obtain the first differential feature parameters of the user, and based on the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, the present invention realizes the update of the facial features of the user over time, so that the digital human model can be continuously adjusted and optimized according to the actual state of the user; obtaining the second differential feature parameters of the user, constructing a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, performing user facial change recognition and matching, and marking and warning users with difficult matching, which helps users more intuitively understand the change trend of the facial state, and realizes the accurate identification and dynamic tracking of the facial health and aging characteristics of the user. Brief Description of the Drawings

[0025] Figure 1 It is a schematic flow chart of an AI digital human modeling method based on user portraits of the present invention.

[0026] Figure 2 It is a schematic structural diagram of an AI digital human modeling system based on user portraits of the present invention. Detailed Embodiments

[0027] The following will clearly and completely describe the technical solutions in the present invention in conjunction with the accompanying drawings in the present invention. Obviously, the described technical solutions are only a part of the present invention, rather than all of it. Based on the technical solutions in the present invention, all other technical solutions obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0028] Figure 1 The flowchart of a method for AI digital human modeling based on user portraits provided in Embodiment 1 of the present invention is shown. As Figure 1 shown, a method for AI digital human modeling based on user portraits in this embodiment includes the following steps:

[0029] Step 1: Obtain the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, perform recognition processing on the original two-dimensional data information to obtain the first facial feature parameters of the user, and obtain the first digital feature parameters based on the original digital human facial information;

[0030] Step 2: Construct a first-level differential feature analysis model based on the first facial features and the first digital feature parameters to obtain the first differential feature parameters of the user; the first differential feature parameters include the first facial aging feature parameters and the first health feature parameters;

[0031] Step 3: Obtain the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, process the first two-dimensional facial data information and the first digital human facial information respectively to obtain the second differential feature parameters of the user;

[0032] Step 4: Construct a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, perform user facial change recognition and matching, and mark and give early warnings to users with difficult matching.

[0033] It should be noted that the first two-dimensional facial data information refers to the user's two-dimensional facial image obtained after obtaining the original two-dimensional facial data information.

[0034] Using the user's two-dimensional facial data information, multiple facial structural parameters are calculated, and the first facial aging characteristic parameter and the first health characteristic parameter are generated. The geometric structure, symmetry, and pixel distribution of the face are analyzed, and the aging characteristics and health status are quantified. Not only is a digital human model generated based on the initial two-dimensional image, but two-dimensional facial data is also re-collected at subsequent time nodes, and new digital human information is generated. By comparing the first difference characteristic parameter and the second difference characteristic parameter, this method realizes the update of the user's facial characteristics over time, so that the digital human model can be continuously adjusted and optimized according to the user's actual state. Through dynamic update, the facial state of the user can be continuously monitored, enabling the system to timely identify significant changes and provide real-time health status feedback to the user. The facial feature change matching model constructed based on the first difference characteristic parameter and the second difference characteristic parameter can compare the feature differences between two time nodes, quantify the facial changes of the user through the user facial matching index, accurately identify the significant change areas, and help the user more intuitively understand the change trend of the facial state. For users with difficult matching, the system can automatically mark and give early warnings, prompting to re-collect data or further adjust the model to ensure the accuracy and applicability of the model. The feature parameters of the two-dimensional image (such as pixel distribution, structural features) are obtained, and the geometric parameters of the three-dimensional digital human model (such as facial curvature, symmetry) are also obtained, realizing multi-modal data fusion.

[0035] Specifically, in step 1, the original two-dimensional facial data information of the user and the original digital human information generated based on the two-dimensional facial data information are obtained. The original two-dimensional data information is identified and processed to obtain the first facial feature parameter of the user. The first facial feature parameter includes the pixel value, the first distribution coefficient, and the first feature mean coefficient of the original two-dimensional facial data information: The frontal two-dimensional facial image of the user in the original two-dimensional facial data information is obtained, and the frontal two-dimensional facial image of the user is segmented. The frontal two-dimensional facial image is segmented into several facial feature regions. In each facial feature region, several feature points are selected. The variance of the pixel values of all feature points is calculated to obtain the first distribution coefficient of each facial feature region, and the average value is calculated by summing the pixel values of all feature points to obtain the first feature mean coefficient.

[0036] Specifically, in step 1, the first digital feature parameter is obtained based on the original digital human facial information: The original digital human facial information is obtained, and the three-dimensional coordinates of the key points of the original digital human facial information are located, that is, the three-dimensional coordinates of the user's eye corners, the three-dimensional coordinates of the nose tip, and the three-dimensional coordinates of the mouth corners. The distances between the two eye corners of the user, the distances between the left and right eye corners and the nose tip, and the distances between the left and right mouth corners and the nose tip are respectively calculated as the first digital feature parameter.

[0037] The calculation formula for the first digital feature parameter is:

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] ;

[0043] Wherein: is the distance between the inner corners of the user's two eyes, is the distance between the left inner corner of the user's eye and the tip of the nose, is the distance between the right inner corner of the user's eye and the tip of the nose, is the distance between the left corner of the user's mouth and the tip of the nose, is the distance between the right corner of the user's mouth and the tip of the nose, is the three-dimensional coordinate of the left inner corner of the user's eye, is the three-dimensional coordinate of the right inner corner of the user's eye, is the three-dimensional coordinate of the tip of the user's nose, is the three-dimensional coordinate of the left corner of the user's mouth, is the three-dimensional coordinate of the right corner of the user's mouth.

[0044] By performing region segmentation and pixel feature extraction on the two-dimensional facial image, the independent analysis of the features of each facial region is ensured. By calculating the variance (the first distribution coefficient) and the average value (the first feature mean coefficient) of the pixel values, the texture features of the facial region (such as wrinkles and skin texture smoothness) are quantified; the variance reflects the change range of the pixels within the region and captures the complexity of the facial texture (such as wrinkle depth), and the average value reflects the overall brightness of the region, helping to identify the skin color distribution and gloss degree; by comparing the distances between the left and right inner corners of the eyes and the tip of the nose and the distances between the left and right corners of the mouth and the tip of the nose, the symmetry of the upper and lower parts of the face can be quantified, helping to detect symmetry changes in the face caused by aging or health problems. Calculating the distance between the inner corners of the eyes and combining the distances between the corners of the eyes and the tip of the nose and the distances between the corners of the mouth and the tip of the nose can provide a quantitative analysis of the facial proportion relationship.

[0045] Specifically, a first-level discrimination feature analysis model is constructed based on the first facial feature and the first digital feature parameter to obtain the first discrimination feature parameter of the user. The formula of the first-level discrimination feature analysis model is:

[0046] ;

[0047] ;

[0048] Wherein: is the first facial aging feature parameter, is the first health feature parameter, is the distance between the inner corners of the user's eyes, is the distance between the left inner corner of the user's eye and the tip of the nose, is the distance between the right inner corner of the user's eye and the tip of the nose, is the distance between the left corner of the user's mouth and the tip of the nose, is the distance between the right corner of the user's mouth and the tip of the nose, is the pixel value of the i-th point in the original two-dimensional facial data information, is the first distribution coefficient, is the first characteristic mean coefficient, is the number of pixel points in the original two-dimensional facial data information;

[0049] Comprehensively considering the changes in the geometric proportions of the user's face, accurately quantifying the changes in facial symmetry, through the relative positions of the key points of the eye corners, the tip of the nose and the corners of the mouth, it can capture significant features such as the decline in symmetry and facial relaxation during the facial aging process. Through the change trend of the geometric parameters in the formula, it can objectively evaluate the degree of facial aging, such as the formation of eye bags, the deepening of nasolabial folds or facial collapse, which helps to identify small but important feature changes, especially the early aging signs of symmetry and relaxation; by analyzing the differences between pixel points and distribution coefficients, it can detect skin texture abnormalities, such as a decrease in smoothness, pigmentation or abnormal freckle distribution, and can sensitively capture subtle health feature changes, making the system have higher recognition accuracy.

[0050] It should be noted that in step 3, obtaining the user's first two-dimensional facial data information and the first digital human facial information generated based on the first two-dimensional facial data information, respectively processing the first two-dimensional facial data information and the first digital human facial information to obtain the user's second discrimination feature parameters: obtaining the user's first two-dimensional facial data information and the first digital human facial information generated based on the first two-dimensional facial data information, performing recognition processing on the first two-dimensional facial data information to obtain the user's second facial feature parameters, and obtaining the second digital feature parameters based on the first digital human facial information; constructing a secondary discrimination feature analysis model based on the second facial features and the second digital feature parameters to obtain the user's second discrimination feature parameters; the second discrimination feature parameters include the second facial aging feature parameters and the second health feature parameters; the formula of the secondary discrimination feature analysis model is:

[0051] ;

[0052] ;

[0053] In the formula: is the second facial aging feature parameter, is the second health feature parameter, is the second distance between the inner corners of the user's eyes, is the second distance between the left inner corner of the user's eye and the tip of the nose, is the second distance between the user's right eye corner and the tip of the nose, is the second distance between the user's left mouth corner and the tip of the nose, is the pixel value of the i-th point in the first two-dimensional facial data information, is the second distribution coefficient, is the second feature mean coefficient, is the number of pixel points in the first two-dimensional facial data information.

[0054] Specifically, the second facial feature parameter includes the pixel value of the first two-dimensional facial data information, the second distribution coefficient, and the second feature mean coefficient: Obtain the user's frontal two-dimensional facial image in the first two-dimensional facial data information, perform segmentation processing on the user's frontal two-dimensional facial image, segment the frontal two-dimensional facial image into several facial feature regions, select several feature points in each facial feature region, calculate the variance of the pixel values of all feature points to obtain the second distribution coefficient of each facial feature region, and calculate the average value of the sum of the pixel values of all feature points to obtain the second feature mean coefficient.

[0055] Obtain the second digital feature parameter based on the first digital human facial information: Obtain the first digital human facial information, and locate the three-dimensional coordinates of the key points of the first digital human facial information, that is, the three-dimensional coordinates of the user's eye corners, the tip of the nose, and the mouth corners, and calculate and obtain the second distances between the two eye corners of the user, the second distances between the left and right eye corners and the tip of the nose, and the second distances between the left and right mouth corners and the tip of the nose as the second digital feature parameters.

[0056] The calculation formula of the second digital feature parameter is:

[0057] ;

[0058] ;

[0059] ;

[0060] ;

[0061] ;

[0062] In the formula: is the second distance between the two eye corners of the user, is the second distance between the user's left eye corner and the tip of the nose, is the second distance between the user's right eye corner and the tip of the nose, is the second distance between the user's left mouth corner and the tip of the nose, is the second distance between the user's right mouth corner and the tip of the nose, is the three-dimensional coordinate of the left eye corner of the first digital human facial information, is the three-dimensional coordinate of the right eye corner of the first digital human face information, is the three-dimensional coordinate of the tip of the nose of the first digital human face information, is the three-dimensional coordinate of the left corner of the mouth of the first digital human face information, is the three-dimensional coordinate of the right corner of the mouth of the first digital human face information.

[0063] The combination of two-dimensional pixel features and three-dimensional space coordinate features makes the description of facial features more comprehensive, capturing both texture information and geometric structure changes; calculating the variance and mean of pixel values within each facial feature area effectively quantifies the uniformity of facial texture and the differences between regions, enhancing the accuracy of feature extraction. The geometric distance calculation based on three-dimensional coordinates can reflect the dynamic changes in the geometric structure of the user's face; regularly updating the user's two-dimensional facial information and the three-dimensional key point data of the digital human can continuously track the change trend of facial features. The dynamic changes in the second distribution coefficient (regional variance) and the second feature mean coefficient can refine the degree of texture aging in specific regions; the division of facial feature areas (such as eyes, nose, mouth) facilitates optimizing the model by region and avoiding the limitations of single features; the second distribution coefficient can quantify the unevenness of texture and brightness distribution within a region, indirectly reflecting the changes in skin texture during the aging process. The second feature mean coefficient provides the overall texture brightness information within a region, and combined with medical data, health problems can be analyzed. The distances between key points (such as the distance between eyes, the distance between the corner of the mouth and the tip of the nose) reflect the change trend of the facial bone and soft tissue structure, helping to evaluate facial structural changes (such as facial fat loss).

[0064] Specifically, in step 4, a facial feature change matching model is constructed based on the first discrimination feature parameter and the second discrimination feature parameter to obtain the user's facial matching index, perform user facial change recognition and matching, and mark and warn users with difficult matching: Obtain the first facial aging feature parameter and the first health feature parameter in the first discrimination feature parameter, and the second facial aging feature parameter and the second health feature parameter in the second discrimination feature parameter. Construct a facial feature change matching model based on the first discrimination feature parameter and the second discrimination feature parameter to obtain the user's facial matching index. The formula of the facial feature change matching model is:

[0065] ;

[0066] In the formula: is the user's facial matching index, is the first facial aging feature parameter, is the first health feature parameter, is the second facial aging feature parameter, is the second health feature parameter;

[0067] Obtain the user's facial matching metrics, compare the user's facial matching metrics with a preset facial matching threshold. If the user's facial matching metrics are greater than or equal to the preset facial matching threshold, mark the user's facial image as difficult to recognize; if the user's facial matching metrics are less than the preset facial matching threshold, there is no need to mark the user's facial image as difficult to recognize.

[0068] Incorporate both facial aging features and health features into the calculation of the matching metrics, comprehensively reflecting the overall changes in the user's face. By comparing the change amounts of two independent parts, the accurate quantification of the change trend is ensured; Obtain the user's facial matching metrics, compare the user's facial matching metrics with a preset facial matching threshold. If the user's facial matching metrics are greater than or equal to the preset facial matching threshold, mark the user's facial image as difficult to recognize; if the user's facial matching metrics are less than the preset facial matching threshold, there is no need to mark the user's facial image as difficult to recognize; Incorporate the facial feature differences (the first discriminative feature parameter and the second discriminative feature parameter) at different time points into the matching model, which can dynamically reflect the facial changes of the user, provide the change trend of the user's facial health and aging, and lay a foundation for long-term facial state management.

[0069] Figure 2The figure shows a schematic structural diagram of an AI digital human modeling system based on user portraits provided in the second embodiment of the present invention. An AI digital human modeling system based on user portraits in this embodiment is used to implement the above-mentioned AI digital human modeling method based on user portraits, and includes a processor and a data acquisition and feature extraction module, a first differential feature parameter analysis module, a second differential feature parameter analysis module, a facial feature change matching evaluation module, and a matching warning module that are communicatively connected to the processor; the data acquisition and feature extraction module is connected to the first differential feature parameter analysis module, the first differential feature parameter analysis module and the second differential feature parameter analysis module are respectively connected to the facial feature change matching evaluation module, and the facial feature change matching evaluation module is connected to the matching warning module; the data acquisition and feature extraction module is used to obtain the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, perform identification processing on the original two-dimensional data information, obtain the first facial feature parameters of the user, and obtain the first digital feature parameters based on the original digital human facial information; the first differential feature parameter analysis module constructs a first-level differential feature analysis model based on the first facial feature and the first digital feature parameters to obtain the first differential feature parameters of the user; the first differential feature parameter analysis module is used to obtain the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, process the first two-dimensional facial data information and the first digital human facial information respectively, and obtain the second differential feature parameters of the user; the facial feature change matching evaluation module is used to construct a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, and perform user facial change recognition and matching; the matching warning module is used to mark and warn users with difficult matching.

[0070] In the embodiment of the present invention, by obtaining the first facial feature and the first digital feature parameters of the user, constructing a first-level differential feature analysis model to obtain the first differential feature parameters of the user, and based on the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, the facial feature update of the user over time is realized, so that the digital human model can be continuously adjusted and optimized according to the actual state of the user; obtaining the second differential feature parameters of the user, constructing a facial feature change matching model according to the first differential feature parameters and the second differential feature parameters to obtain the user facial matching index, performing user facial change recognition and matching, and marking and warning users with difficult matching, which helps users more intuitively understand the change trend of the facial state, and realizes the accurate identification and dynamic tracking of the facial health and aging characteristics of the user.

[0071] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

[0072] Finally: The above description is only the preferred solution of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An AI digital human modeling method based on user portrait, characterized in that: The steps include: Step 1, obtaining the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, performing recognition processing on the original two-dimensional data information, obtaining the first facial feature parameter of the user, and obtaining the first digital feature parameter based on the original digital human facial information; Step 2: construct a first-level distinguishing feature analysis model based on the first facial feature and the first digital feature parameter to obtain the first distinguishing feature parameter of the user; the first distinguishing feature parameter includes a first facial aging feature parameter and a first health feature parameter; the formula of the first-level distinguishing feature analysis model is: ； ； Where: is the first facial aging characteristic parameter, is the first health characteristic parameter, is the distance between the user's two eye corners, The distance between the user's left eye corner and nose tip, is the distance between the user's right eye corner and nose tip, is the distance between the user's left mouth corner and nose tip, is the distance between the user's right mouth corner and nose tip, is the pixel value of the i-th point in the original two-dimensional facial data information, is the first distribution coefficient, is the first eigenvalue coefficient, is the number of pixels in the original two-dimensional facial data information; Step 3, obtaining the user's first two-dimensional facial data information and the first digital human facial information generated based on the first two-dimensional facial data information, and processing the first two-dimensional facial data information and the first digital human facial information respectively to obtain the user's second distinguishing feature parameter; Step 4: construct a facial feature change matching model based on the first distinguishing feature parameter and the second distinguishing feature parameter to obtain the user facial matching index, perform user facial change recognition and matching, and mark and warn users who are difficult to match.

2. The AI ​​digital human modeling method based on user portrait according to claim 1 is characterized in that: In step 1, the original two-dimensional facial data information of the user and the original digital human information generated based on the two-dimensional facial data information are obtained, the original two-dimensional data information is identified and processed, and the first facial feature parameters of the user are obtained, and the first facial feature parameters include the pixel value of the original two-dimensional facial data information, the first distribution coefficient and the first feature mean coefficient: the front two-dimensional facial image of the user in the original two-dimensional facial data information is obtained, the front two-dimensional facial image of the user is segmented, and the front two-dimensional facial image is segmented into a number of facial feature areas, a number of feature points are selected in each facial feature area, the pixel values ​​of all feature points are calculated by variance to obtain the first distribution coefficient of each facial feature area, and the pixel values ​​of all feature points are summed and the average value is calculated to obtain the first feature mean coefficient.

3. The AI ​​digital human modeling method based on user portrait according to claim 1 is characterized in that: In step 1, the first digital feature parameter is obtained based on the original digital human facial information: the original digital human facial information is obtained, and the three-dimensional coordinates of the key points of the original digital human facial information are located, namely, the three-dimensional coordinates of the user's eye corners, the three-dimensional coordinates of the nose tip, and the three-dimensional coordinates of the mouth corners, and the distance between the user's two eye corners, the distance between the left and right eye corners and the nose tip, and the distance between the left and right mouth corners and the nose tip are respectively calculated and obtained as the first digital feature parameter.

4. The AI ​​digital human modeling method based on user portrait according to claim 1 is characterized in that: In step 3, the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information are obtained, and the first two-dimensional facial data information and the first digital human facial information are processed respectively to obtain the second distinguishing feature parameters of the user: the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information are obtained, the first two-dimensional facial data information is identified and processed to obtain the second facial feature parameters of the user, and the second digital feature parameters are obtained based on the first digital human facial information; a secondary distinguishing feature analysis model is constructed based on the second facial features and the second digital feature parameters to obtain the second distinguishing feature parameters of the user; the second distinguishing feature parameters include a second facial aging feature parameter and a second health feature parameter; the formula of the secondary distinguishing feature analysis model is: ； ； Where: is the second facial aging characteristic parameter, is the second health characteristic parameter, is the second distance between the two corners of the user's eyes, The second distance between the user's left eye corner and nose tip, is the second distance between the user's right eye corner and nose tip, is the second distance between the user's left mouth corner and nose tip, is the pixel value of the i-th point in the first two-dimensional facial data information, is the second distribution coefficient, is the second characteristic mean coefficient, is the number of pixels in the first two-dimensional facial data information.

5. The AI ​​digital human modeling method based on user portrait according to claim 1 is characterized in that: Step 4, constructing a facial feature change matching model based on the first distinguishing feature parameter and the second distinguishing feature parameter to obtain the user's facial matching index, performing user facial change recognition and matching, and marking and warning users with matching difficulties: obtaining the first facial aging feature parameter and the first health feature parameter in the first distinguishing feature parameter, the second facial aging feature parameter and the second health feature parameter in the second distinguishing feature parameter, constructing a facial feature change matching model based on the first distinguishing feature parameter and the second distinguishing feature parameter to obtain the user's facial matching index, the formula of the facial feature change matching model is: ； Where: is the user's facial matching indicator, is the first facial aging characteristic parameter, is the first health characteristic parameter, is the second facial aging characteristic parameter, is the second health characteristic parameter; Obtain a user facial matching index, and compare the user facial matching index with a preset facial matching threshold. If the user facial matching index is greater than or equal to the preset facial matching threshold, mark the user facial image as difficult to recognize; if the user facial matching index is less than the preset facial matching threshold, there is no need to mark the user facial image as difficult to recognize.

6. An AI digital human modeling system based on user portrait, used to implement an AI digital human modeling method based on user portrait according to any one of claims 1 to 5, characterized in that: It includes a processor and a data acquisition and feature extraction module, a first distinguishing feature parameter analysis module, a second distinguishing feature parameter analysis module, a facial feature change matching evaluation module and a matching warning module connected to the processor in communication; the data acquisition and feature extraction module is used to obtain the original two-dimensional facial data information of the user and the original digital human facial information generated based on the two-dimensional facial data information, identify and process the original two-dimensional data information, obtain the first facial feature parameter of the user, and obtain the first digital feature parameter based on the original digital human facial information; the first distinguishing feature parameter analysis module constructs a first-level distinguishing feature analysis model based on the first facial feature and the first digital feature parameter, and obtains the first distinguishing feature parameter of the user; the first distinguishing feature parameter analysis module is used to obtain the first two-dimensional facial data information of the user and the first digital human facial information generated based on the first two-dimensional facial data information, and processes the first two-dimensional facial data information and the first digital human facial information respectively to obtain the second distinguishing feature parameter of the user; The facial feature change matching assessment module is used to construct a facial feature change matching model based on the first distinguishing feature parameter and the second distinguishing feature parameter to obtain the user facial matching index and perform user facial change recognition and matching; the matching warning module is used to mark and warn users who are difficult to match.

Images