A method for constructing a 3D facial model for facial cosmetic surgery

Abstract

The present invention discloses a 3D facial model construction method for facial cosmetic surgery, relates to the technical field of cosmetic surgery model construction, and improves the scientificity and accuracy of cosmetic surgery decisions. The present invention matches each post-cosmetic surgery image with a user's facial feature points, establishes a facial change prediction network according to the matching results, establishes a user facial model according to facial image data taken at various angles, maps the facial change prediction network onto the user facial model, generates a number of cosmetic surgery displacement vectors on the facial change prediction network according to the facial cosmetic surgery decision, obtains the expected change position of each user's facial feature point according to the cosmetic surgery image associated with each cosmetic surgery displacement vector corresponding to the user's facial feature point, and then synchronously adjusts the facial cosmetic surgery decision. The user facial model is adjusted according to the optimized facial cosmetic surgery decision to obtain a 3D facial cosmetic surgery result model.

Description

Technical Field

[0001] The present invention relates to the technical field of cosmetic model construction, and in particular to a method for constructing a 3D facial model for facial cosmetic surgery. Background Art

[0002] In the field of cosmetic surgery, traditional methods for predicting surgical results often have numerous limitations. Traditionally, doctors relied solely on their own experience and limited two-dimensional images to describe the general effects of surgery. This approach made it difficult for patients to intuitively and accurately perceive the specific changes from various angles after surgery. This led to significant discrepancies between patient expectations and actual results, impacting patient satisfaction.

[0003] Furthermore, traditional methods lack accurate analysis of individual facial features and personalized effect simulation. Facial features vary significantly between patients, and simply referencing generic plastic surgery cases cannot meet the specific needs of each patient. Furthermore, communication between doctors and patients during the plastic surgery decision-making process is not efficient and accurate, making it difficult to adjust the plastic surgery plan in real time according to the patient's expectations. This can result in the final plastic surgery plan not being the most suitable for the patient.

[0004] While some facial simulation technologies have emerged with the advancement of technology, most of these technologies can only perform simple two-dimensional simulations and cannot provide a realistic and comprehensive three-dimensional display of plastic surgery results. They also struggle to dynamically and accurately adjust the model based on the patient's specific facial features and plastic surgery needs. Therefore, a method for constructing a 3D facial model for facial cosmetic surgery is provided. Summary of the Invention

[0005] In order to solve the above technical problems, the purpose of the present invention is to provide a 3D facial model construction method for facial cosmetic surgery.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for constructing a 3D facial model for facial cosmetic surgery comprises the following steps:

[0008] Step S1: photographing the user's face from multiple angles to generate facial image data, matching the facial image data taken from various angles via the Internet, and then obtaining a number of images after plastic surgery;

[0009] Step S2: extracting a plurality of user facial feature points from the facial image data taken at various angles, matching each post-plastic surgery image with the user's facial feature points, and associating each post-plastic surgery image with the user's facial feature points based on the matching results, thereby establishing a facial change prediction network, establishing a user facial model based on the facial image data taken at various angles, and mapping the facial change prediction network onto the user facial model;

[0010] Step S3: Obtaining a facial cosmetic surgery decision, and generating a plurality of cosmetic surgery displacement vectors on a facial change prediction network based on the facial cosmetic surgery decision. Based on the post-cosmetic surgery image associated with each facial feature point of the user corresponding to each cosmetic surgery displacement vector, the predicted change position of each facial feature point of the user is obtained, and the facial cosmetic surgery decision is then adjusted synchronously.

[0011] Step S4: reposition the user's facial feature points on the user's facial model according to the optimized facial cosmetic surgery decision, and synchronously adjust the entire user's facial model, thereby outputting a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision.

[0012] Furthermore, the facial image data collection process includes:

[0013] Deploy 12 high-resolution cameras in a circular array in a standard studio. Each camera is evenly distributed around the user's head at 15-degree intervals, ensuring that the camera's capture range covers the user's face from the front, side, top, and bottom perspectives. Furthermore, the capture ranges of adjacent cameras overlap.

[0014] All high-resolution cameras are triggered synchronously to shoot and obtain facial image data of the user's face under natural light and specific wavelengths.

[0015] Furthermore, the process of obtaining the post-plastic surgery image includes:

[0016] Using a deep hash-based image retrieval algorithm, with facial image data taken from various angles as index material, it quickly matches post-surgery images with a similarity of more than 90% through the internet;

[0017] After the matching of the post-plastic surgery images is completed, a two-dimensional coordinate system is established, and the post-plastic surgery images and facial image data are overlapped and mapped to the two-dimensional coordinate system. Then, by using affine transformation, the key points of the post-plastic surgery images are aligned with the key points of the user's current face to eliminate the deviation caused by the difference in shooting angles.

[0018] Furthermore, the process of extracting a number of user facial feature points from facial image data captured at various angles includes:

[0019] Matching and connecting facial image data taken at various angles to obtain user facial image data, and establishing a user facial model based on the user facial image data;

[0020] A cascaded convolutional neural network is used to extract n user facial feature points from facial image data taken at various angles, where n is a natural number greater than 0;

[0021] The facial feature points of the user on the facial image data taken at various angles are matched with each other. If it is determined that two facial feature points of the user correspond to the same position, the corresponding facial feature points of the user are merged, otherwise no operation is performed.

[0022] Furthermore, the process of matching the post-surgery image with the user's facial feature points includes:

[0023] Extract a number of front facial feature points and back facial feature points from the images before and after plastic surgery, and use each user's facial feature point as the central feature point in turn, and compare the front facial feature points and the central feature points on each of the images before and after plastic surgery. The comparison process includes:

[0024] Taking the central feature point as the center, select the nearest user facial feature point every 10° in the 360° direction as the associated user facial feature point. Then, match the front facial feature points on each pre- and post-surgery image with the central feature point. Based on the matching results, obtain the spatial distance between each front facial feature point and the central feature point.

[0025] Set a spatial distance threshold. If the spatial distance between the front facial feature point and the center feature point is less than or equal to the spatial distance threshold, then select the associated user facial feature points based on the process of selecting the center feature point, and select the associated front facial feature points around the corresponding front facial feature point.

[0026] If the spatial distance between the front feature point and the center feature point is greater than the spatial distance threshold, no operation is performed.

[0027] Furthermore, the process of establishing the facial change prediction network includes:

[0028] Establishing a plurality of user face extension vectors and front face extension vectors with the central feature point and the front face feature point as the starting point and the associated user face feature points and the associated front face feature points as the end point, matching the user face extension vectors and the front face extension vectors with each other, and then obtaining the product of the user face extension vector and the front face extension vector with the smallest angle between the two;

[0029] The total number of products between each user's facial extension vector and the front facial extension vector is accumulated and recorded as the matching correlation degree. A correlation degree threshold is set. If there is a matching correlation degree between a front facial feature point and a central feature point that is greater than or equal to the correlation degree threshold, the pre- and post-surgery images of the front facial feature point are correlated with the corresponding central feature point. Otherwise, no operation is performed.

[0030] Each user's facial feature point is mapped to the user's facial model, and based on the results of matching the central feature point with the associated user's facial feature points, each user's facial feature point is connected with its matched associated user's facial feature points, thereby obtaining a facial change prediction network, and the pre- and post-surgery images matched by each user's facial feature point are synchronously associated with the facial change prediction network.

[0031] Furthermore, the process of generating a plurality of cosmetic surgery displacement vectors on the facial change prediction network according to the facial cosmetic surgery decision includes:

[0032] The facial cosmetic surgery decision includes the estimated change position of each part on the user's facial model, establishes a three-dimensional coordinate system, binds the estimated change position of each part in the facial cosmetic surgery decision to the user's facial feature points on the user's facial model and maps them into the three-dimensional coordinate system, obtains the estimated spatial displacement position of each user's facial feature point based on the facial cosmetic surgery decision, and then generates a multidimensional vector D corresponding to the facial cosmetic surgery decision;

[0033] The multidimensional vector D={d1, d2, ······, d n}, where d n Represents the cosmetic displacement vector of the nth user's facial feature point.

[0034] Further, the process of adjusting facial cosmetic surgery decisions includes:

[0035] Establish an optimization objective function that takes into account both aesthetic score and physiological feasibility:

[0036]

[0037] in and Represent the aesthetic loss of the golden ratio and the muscle and bone constraint loss, and represents the constant correction parameter, Indicates the coordinated value;

[0038] According to the post-plastic surgery image associated with each user's facial feature point, a number of associated displacement vectors are set for the cosmetic surgery displacement vector of each user's facial feature point in the multidimensional vector D. The angle between each associated displacement vector and the corresponding cosmetic surgery displacement vector is (-90°, 90°), and the angle between each associated displacement vector and the corresponding cosmetic surgery displacement vector is different. At the same time, the starting point of the associated displacement vector is the user's facial feature point, but the end position is on the post-plastic surgery image;

[0039] That is, each associated displacement vector comes from the front-to-back position change of the position associated with the user's facial feature points in the post-plastic surgery image;

[0040] Input each associated displacement vector and cosmetic displacement vector into the optimization objective function, and then optimize the objective function to output the associated displacement vector or cosmetic displacement vector with the smallest coordination value, and record the output displacement vector as the expected change position of the corresponding user's facial feature point;

[0041] When the estimated change positions of all the user's facial feature points are updated, the correction of the facial cosmetic surgery decision is completed.

[0042] Furthermore, the generation process of the 3D facial plastic surgery result model includes:

[0043] The facial change prediction network maps the expected change position of each user's facial feature points within the facial cosmetic surgery decision to the user's facial model, and then uses the Laplace smoothing algorithm to eliminate the deformation of the user's facial model caused by the change in the position of the user's facial feature points:

[0044] After the position changes and corrections of all the user's facial feature points are completed according to the facial cosmetic surgery decision, a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision is output.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] 1. Extract the user's facial feature points from facial image data, match and associate them with the post-surgery image, build a facial change prediction network, and map it to the user's facial model. This approach fully considers each user's unique facial features and can provide highly personalized plastic surgery effect predictions to meet the specific needs of different users.

[0047] 2. This invention generates and adjusts a cosmetic surgery displacement vector based on the facial cosmetic surgery decision, simultaneously adjusting the decision. This enables dynamic adjustments based on the user's facial features and expectations in real time during the cosmetic surgery decision-making process, ensuring the final cosmetic surgery plan is the most suitable for the user, improving the scientific nature and accuracy of cosmetic surgery decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention.

[0049] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0050] To make the objectives, technical solutions, and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementations obtained by those of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0051] like Figure 1 As shown, a method for constructing a 3D facial model for facial cosmetic surgery includes the following steps:

[0052] Step S1: photographing the user's face from multiple angles to generate facial image data, matching the facial image data taken from various angles via the Internet, and then obtaining a number of images after plastic surgery;

[0053] Step S2: extracting a plurality of user facial feature points from the facial image data taken at various angles, matching each post-plastic surgery image with the user's facial feature points, and associating each post-plastic surgery image with the user's facial feature points based on the matching results, thereby establishing a facial change prediction network, establishing a user facial model based on the facial image data taken at various angles, and mapping the facial change prediction network onto the user facial model;

[0054] Step S3: Obtaining a facial cosmetic surgery decision, and generating a plurality of cosmetic surgery displacement vectors on a facial change prediction network based on the facial cosmetic surgery decision. Based on the post-cosmetic surgery image associated with each facial feature point of the user corresponding to each cosmetic surgery displacement vector, the predicted change position of each facial feature point of the user is obtained, and the facial cosmetic surgery decision is then adjusted synchronously.

[0055] Step S4: reposition the user's facial feature points on the user's facial model according to the optimized facial cosmetic surgery decision, and synchronously adjust the entire user's facial model, thereby outputting a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision.

[0056] Furthermore, step S1 is implemented by the following steps:

[0057] Deploy a circular array of 12 high-resolution cameras (with a resolution of no less than 24 megapixels) in a standard studio. Each high-resolution camera is evenly distributed around the user's head at 15-degree intervals, ensuring that the high-resolution camera's capture range covers the user's face from the front, side, top, and bottom perspectives. Furthermore, the capture ranges of high-resolution cameras positioned adjacently in sequential spatial locations partially overlap.

[0058] Synchronously triggering all high-resolution cameras to capture facial image data of the user's face under natural light and specific wavelengths (such as UV light). The facial image data includes facial structural features such as the user's skin texture, pore distribution, and subcutaneous blood vessels;

[0059] Using a deep hash-based image retrieval algorithm, with facial image data taken from various angles as index material, we quickly match images of cosmetic surgery with a similarity higher than 90% through the Internet. The hash algorithm formula is:

[0060] ;

[0061] in represents the image feature analysis result, W is the hash weight matrix, I represents the facial image data, and Represent the pooling operation and convolution operation respectively, and T represents the matrix dimension;

[0062] After the matching of the post-plastic surgery images is completed, a two-dimensional coordinate system is established, and the post-plastic surgery images and facial image data are overlapped and mapped to the two-dimensional coordinate system. Then, by using affine transformation, the key points of the post-plastic surgery images (such as the corners of the eyes and the tip of the nose) are aligned with the key points of the user's current face to eliminate the deviation caused by the difference in shooting angles.

[0063] Furthermore, step S2 is implemented by the following steps:

[0064] The facial image data taken at various angles are matched and connected. Since each facial image data corresponds to the same user, there must be some common parts in the facial image data. Then, the facial image data are stitched together to obtain the user's facial image data, and a user facial model is established based on the user's facial image data.

[0065] A cascaded convolutional neural network is used to extract n user facial feature points from facial image data taken at various angles. The user facial feature points include anatomical landmarks such as brow arches, eyelids, nose wings, and mouth corners. The extraction formula is:

[0066] ;

[0067] in represents the facial feature points of the i-th user, Indicates the shooting angle, represents the convolution operation, I represents the facial image data, i is a natural number less than or equal to n, and n is a natural number greater than 0;

[0068] Match the facial feature points of the user on the facial image data taken at various angles. If two facial feature points of the user correspond to the same position, merge the corresponding facial feature points; otherwise, do nothing.

[0069] The process of extracting user facial feature points from facial image data is used to extract a number of front facial feature points and back facial feature points from the images before and after plastic surgery. Each user facial feature point is used as a central feature point, and the front facial feature points on each of the images before and after plastic surgery are compared with the central feature points. The comparison process includes:

[0070] Taking the central feature point as the center, select the nearest user facial feature point every 10° in the 360° direction as the associated user facial feature point. Then, match the front facial feature points on each pre- and post-surgery image with the central feature point. Based on the matching results, obtain the spatial distance between each front facial feature point and the central feature point.

[0071] Set a spatial distance threshold. If the spatial distance between the front facial feature point and the center feature point is less than or equal to the spatial distance threshold, then select the associated user facial feature points based on the process of selecting the center feature point, and select the associated front facial feature points around the corresponding front facial feature point.

[0072] If the spatial distance between the front feature point and the center feature point is greater than the spatial distance threshold, no operation is performed;

[0073] Establishing a plurality of user face extension vectors and front face extension vectors with the central feature point and the front face feature point as the starting point and the associated user face feature points and the associated front face feature points as the end point, matching the user face extension vectors and the front face extension vectors with each other, and then obtaining the product of the user face extension vector and the front face extension vector with the smallest angle between the two;

[0074] The total number of products between each user's facial extension vector and the front facial extension vector is accumulated and recorded as the matching correlation degree. A correlation degree threshold is set. If there is a matching correlation degree between a front facial feature point and a central feature point that is greater than or equal to the correlation degree threshold, the pre- and post-surgery images of the front facial feature point are correlated with the corresponding central feature point. Otherwise, no operation is performed.

[0075] Each user's facial feature point is mapped to the user's facial model, and based on the results of matching the central feature point with the associated user's facial feature points, each user's facial feature point is connected with its matched associated user's facial feature points, thereby obtaining a facial change prediction network, and the pre- and post-surgery images matched by each user's facial feature point are synchronously associated with the facial change prediction network.

[0076] Furthermore, step S3 is implemented by the following steps:

[0077] The doctor uploads a facial cosmetic surgery decision, which includes the expected changes to various parts of the user's facial model, such as raising the nose bridge, reducing the mandibular angle, etc.

[0078] Establish a three-dimensional coordinate system, bind the predicted change position of each part in the facial cosmetic surgery decision to the user's facial feature points on the user's facial model and map them into the three-dimensional coordinate system, obtain the predicted spatial displacement position of each user's facial feature point based on the facial cosmetic surgery decision, and then generate a multidimensional vector D corresponding to the facial cosmetic surgery decision;

[0079] The multidimensional vector D={d1, d2, ······, d n}, where d n Represents the cosmetic displacement vector of the nth user's facial feature point, n≥1;

[0080] Establish an optimization objective function that takes into account both aesthetic score and physiological feasibility:

[0081]

[0082] in and Represent the aesthetic loss of the golden ratio and the muscle and bone constraint loss, and represents the constant correction parameter, Indicates the coordinated value;

[0083] According to the post-plastic surgery image associated with each user's facial feature point, m associated displacement vectors are set for the cosmetic surgery displacement vector of each user's facial feature point in the multidimensional vector D. The angle between each associated displacement vector and the corresponding cosmetic surgery displacement vector is (-90°, 90°), and the angle between each associated displacement vector and the corresponding cosmetic surgery displacement vector is different. At the same time, the starting point of the associated displacement vector is the user's facial feature point, but the end position is on the post-plastic surgery image. m is a natural number greater than 0;

[0084] That is, each associated displacement vector comes from the front-to-back position change of the position associated with the user's facial feature points in the post-plastic surgery image;

[0085] Input each associated displacement vector and cosmetic displacement vector into the optimization objective function, and then optimize the objective function to output the associated displacement vector or cosmetic displacement vector with the smallest coordination value, and record the output displacement vector as the expected change position of the corresponding user's facial feature point;

[0086] When the estimated change positions of all the user's facial feature points are updated, the correction of the facial cosmetic surgery decision is completed.

[0087] Furthermore, step S4 is implemented by the following steps:

[0088] The facial change prediction network maps the expected position changes of each user's facial feature points within the facial cosmetic surgery decision to the user's facial model, and then uses the Laplace smoothing algorithm to eliminate the deformation of the user's facial model caused by the change in the position of the user's facial feature points:

[0089] ;

[0090] in ) represents the coordinate position of the changed facial feature point of the associated user of the i-th user’s facial feature point, represents the coordinate position of the t,i-th user's facial feature point according to the expected change position in the facial cosmetic surgery decision, represents the corrected coordinates of the facial feature points of the i-th user, is the smoothing strength parameter, ∈(0,1);

[0091] After the position changes and corrections of all the user's facial feature points are completed according to the facial cosmetic surgery decision, a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision is output.

[0092] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for constructing a 3D facial model for facial cosmetic surgery, characterized in that: The following steps are involved: Step S1: photographing the user's face from multiple angles to generate facial image data, matching the facial image data taken from various angles via the Internet, and then obtaining a number of images after plastic surgery; Step S2: extracting a plurality of user facial feature points from the facial image data taken at various angles, matching each post-plastic surgery image with the user's facial feature points, and associating each post-plastic surgery image with the user's facial feature points based on the matching results, thereby establishing a facial change prediction network, establishing a user facial model based on the facial image data taken at various angles, and mapping the facial change prediction network onto the user facial model; The matching process between the post-surgery image and the user's facial feature points includes: Extract a number of front and back facial feature points from the images before and after plastic surgery, and use each user's facial feature point as the central feature point in turn; Taking the central feature point as the center, select the closest user facial feature point every 10° in the 360° direction as the associated user facial feature point. Then, match the front facial feature points on each pre- and post-surgery image with the central feature point. Based on the matching results, obtain the spatial distance between each front facial feature point and the central feature point. According to the relationship between the spatial distance between the front facial feature point and the central feature point and the spatial distance threshold, the associated front facial feature points are selected around the corresponding front facial feature point; The process of establishing the facial change prediction network includes: Establishing a plurality of user face extension vectors and front face extension vectors with the central feature point and the front face feature point as the starting point and the associated user face feature points and the associated front face feature points as the end point, matching the user face extension vectors and the front face extension vectors with each other, and then obtaining the product of the user face extension vector and the front face extension vector with the smallest angle between the two; The total number of products between each user's facial extension vector and the front facial extension vector is accumulated and recorded as the matching correlation degree. If there is a matching correlation degree between a front facial feature point and a central feature point that is greater than or equal to the correlation degree threshold, the pre- and post-surgery images of the front facial feature point are associated with the corresponding central feature point. Connecting each user's facial feature points with the facial feature points of its associated users to obtain a facial change prediction network, and synchronously associating the pre- and post-surgery images of each user's facial feature points with the facial change prediction network; Step S3: Obtaining a facial cosmetic surgery decision, and generating a plurality of cosmetic surgery displacement vectors on a facial change prediction network based on the facial cosmetic surgery decision. Based on the post-cosmetic surgery image associated with each facial feature point of the user corresponding to each cosmetic surgery displacement vector, the predicted change position of each facial feature point of the user is obtained, and the facial cosmetic surgery decision is then adjusted synchronously. Step S4: reposition the user's facial feature points on the user's facial model according to the optimized facial cosmetic surgery decision, and synchronously adjust the entire user's facial model, thereby outputting a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision.

2. A 3D facial model construction method for facial cosmetic surgery according to claim 1, characterized in that: The facial image data collection process includes: High-resolution cameras are deployed in a circular array in a standardized studio. Each high-resolution camera is evenly distributed around the user's head at intervals of 15°, ensuring that the shooting range of the high-resolution camera covers the front, side, top and bottom perspectives of the user's face. At the same time, the shooting ranges of high-resolution cameras distributed in adjacent sequential spatial positions partially overlap, and all high-resolution cameras are synchronously triggered to capture facial image data of the user's face.

3. The method for constructing a 3D facial model for facial cosmetic surgery according to claim 2, wherein: The process of obtaining the image after plastic surgery includes: Using facial image data taken at various angles as index material, the system quickly matches post-plastic surgery images with a similarity of more than 90% through the Internet. Once the post-plastic surgery image matching is completed, a two-dimensional coordinate system is established, and each post-plastic surgery image and facial image data are overlapped and mapped to the two-dimensional coordinate system. Then, by using affine transformation, the key points of the post-plastic surgery image are aligned with the key points of the user's current face, eliminating the deviation caused by differences in shooting angles.

4. The method for constructing a 3D facial model for facial cosmetic surgery according to claim 3, wherein: The process of extracting several user facial feature points from facial image data taken at various angles includes: A user face model is established based on facial image data taken at various angles, and n user facial feature points are extracted from the facial image data taken at various angles using a cascaded convolutional neural network, where n is a natural number greater than 0; The facial feature points of the user on the facial image data taken at various angles are matched with each other. If it is determined that two facial feature points of the user correspond to the same position, the corresponding facial feature points of the user are merged, otherwise no operation is performed.

5. The method for constructing a 3D facial model for facial cosmetic surgery according to claim 4, wherein: The process of generating several cosmetic surgery displacement vectors on the facial change prediction network according to the facial cosmetic surgery decision includes: The facial cosmetic surgery decision includes the expected change position of each part on the user's facial model, establishes a three-dimensional coordinate system, binds the expected change position of each part in the facial cosmetic surgery decision to the user's facial feature points on the user's facial model and maps them into the three-dimensional coordinate system, obtains the expected spatial displacement position of each user's facial feature point based on the facial cosmetic surgery decision, and then generates a multidimensional vector D corresponding to the facial cosmetic surgery decision.

6. The method for constructing a 3D facial model for facial cosmetic surgery according to claim 5, wherein: The process of adjusting a facial cosmetic surgery decision includes: Establishing an optimization objective function, and setting a number of associated displacement vectors for the cosmetic surgery displacement vector of each user's facial feature point in the multidimensional vector D according to the cosmetic surgery image associated with each user's facial feature point; Input each associated displacement vector and cosmetic displacement vector into the optimization objective function, and then optimize the objective function to output the associated displacement vector or cosmetic displacement vector with the smallest coordination value, and record the output displacement vector as the expected change position of the corresponding user's facial feature point; When the estimated change positions of all the user's facial feature points are updated, the correction of the facial cosmetic surgery decision is completed.

7. The method for constructing a 3D facial model for facial cosmetic surgery according to claim 6, wherein: The generation process of the 3D facial plastic surgery result model includes: The facial change prediction network maps the expected change position of each user's facial feature points within the facial cosmetic surgery decision to the user's facial model, and then uses the Laplace smoothing algorithm to eliminate the deformation of the user's facial model caused by the change in the position of the user's facial feature points: After the position changes and corrections of all the user's facial feature points are completed according to the facial cosmetic surgery decision, a 3D facial cosmetic surgery result model corresponding to the facial cosmetic surgery decision is output.

Images