CN113284229B - Three-dimensional face model generation method, device, equipment and storage medium - Google Patents

Abstract

The invention relates to the field of image processing, and discloses a method, a device, equipment and a storage medium for generating a three-dimensional face model, wherein the method comprises the following steps: acquiring face image information of a user, and performing feature recognition on the face image information to acquire face feature information of the user; inputting the face feature information into a preset three-dimensional conversion neural network model for three-dimensional conversion processing to obtain an initial face model of a user; the method comprises the steps of obtaining preset rendering data, performing effect rendering on an initial-order face model based on the preset rendering data to obtain a three-dimensional face model of a user, improving modeling efficiency by performing model matching and model adjustment in a preset three-dimensional conversion neural network model based on obtained face feature information, and performing effect rendering on the obtained initial-order face model based on the preset rendering data to improve personalized display effect of the obtained three-dimensional face model, further improving network social experience of the user and increasing user viscosity.

Description

Three-dimensional face model generation method, device, equipment and storage medium

technical field

The present invention relates to the technical field of image processing, in particular to a method, device, equipment and storage medium for generating a three-dimensional face model.

Background technique

When a user chooses a social APP to socialize, he will upload a photo or an online picture as his avatar (for example, "Upload Avatar" in the "Personal Center"). Other users are the same, but the problem of "colliding head portraits" occurs, and using selfies or Internet pictures as head portraits cannot bring novelty to users, lacks interest, and is difficult to stimulate users' self-expression awareness. In the prior art, although some The manufacturer has added the function of generating a corresponding face model based on the user's real-time selfie in the selfie function, but there are still problems of low accuracy of the face model or low construction efficiency and long construction time of the face model. Therefore, how to improve the personalization The generation efficiency and display effect of the 3D face model, in order to improve the user's online social experience and increase user stickiness, has become an urgent problem to be solved.

The above content is only used to assist in understanding the technical solution of the present invention, and does not mean that the above content is admitted as prior art.

Contents of the invention

The main purpose of the present invention is to provide a method, device, device, and storage medium for generating a three-dimensional face model, aiming to solve how to improve the generation efficiency and display effect of a personalized three-dimensional face model, so as to improve the user's network social experience, Technical issues that increase user stickiness.

To achieve the above object, the invention provides a method for generating a three-dimensional human face model, the method comprising the following steps:

Obtaining the user's face image information, and performing feature recognition on the face image information to obtain the user's face feature information;

inputting the face feature information into a preset three-dimensional transformation neural network model for three-dimensional transformation processing to obtain the user's primary face model;

Acquiring preset rendering data, and performing effect rendering on the primary face model based on the preset rendering data, so as to obtain a three-dimensional face model of the user.

Optionally, the step of acquiring the user's face image information and performing feature recognition on the face image information to obtain the user's face feature information specifically includes:

Acquire the user's face image information, and perform image recognition on the face image information to obtain key points of the face image;

Converting the key points of the human face image into a preset human face coordinate system to obtain the key point coordinate difference between the key points of the human face image and the corresponding preset image key points in the preset human face coordinate system ;

Generate facial feature information of the user based on the key point coordinate difference.

Optionally, the step of inputting the facial feature information into a preset three-dimensional transformation neural network model to perform three-dimensional transformation processing, so as to obtain the user's primary human face model, specifically includes:

Carrying out model matching between the face feature information and the model feature information of each preset face model in the preset three-dimensional conversion neural network model, to obtain the model matching degree between the face feature information and the model feature information;

Sorting the matching degree of the models to obtain a matching degree ranking result, and determining a target face model based on the matching degree ranking result;

Adaptively adjusting the target face model based on the face feature information to obtain a primary face model of the user.

Optionally, the step of sorting the matching degree of the models to obtain the matching degree ranking result, and determining the target face model based on the matching degree ranking result, specifically includes:

Sorting the matching degrees of the models from large to small to obtain the matching degree sorting results;

Judging whether the model matching degree corresponding to the preset face model with the first ranking order in the matching degree sorting result is greater than the preset matching degree;

When the preset matching degree is greater than or equal to the preset face model with the first rank as the target face model;

When it is less than the preset matching degree, the preset standard model is used as the target face model.

Optionally, the step of adaptively adjusting the target face model based on the face feature information to obtain the user's primary face model specifically includes:

Obtaining the face feature values corresponding to the face feature information in each face area, and the target face model corresponding to face target values in each preset partition;

Comparing the feature value of the human face with the target value of the human face to obtain a feature comparison result;

Adaptively adjust each face area of the target face model based on the feature comparison result to obtain a primary face model of the user.

Optionally, the preset rendering data includes preset texture data and preset style data;

Correspondingly, the step of acquiring preset rendering data, and performing effect rendering on the primary face model based on the preset rendering data, so as to obtain the user's 3D face model, specifically includes:

Obtaining the preset texture data and the preset style data corresponding to each face area, and edge point information and normal vector information corresponding to each face area;

Mapping the primary face model based on the preset texture data corresponding to each face area, the edge point information, and the normal vector information to obtain a face texture model;

Render the face texture model based on the preset style data, the edge point information, and the normal vector information corresponding to each face area, to obtain a three-dimensional face model of the user.

Optionally, the step of acquiring preset rendering data, and performing effect rendering on the primary face model based on the preset rendering data, so as to obtain the user's 3D face model, specifically includes:

Acquiring one type of preference information in the adjustment preference information input by the user, and adjusting the primary face model based on the one type of preference information to obtain a user fine-tuning model;

Obtain preset rendering data and the second-type preference information in the adjustment preference information, and perform effect rendering on the user fine-tuned model based on the rendering data and the second-type preference information, so as to obtain the user's three-dimensional face Model.

In addition, in order to achieve the above object, the present invention also proposes a 3D face model generation device, the 3D face model generation device includes:

A feature recognition module, configured to acquire the user's face image information, and perform feature recognition on the face image information to obtain the user's face feature information;

A three-dimensional conversion module, configured to input the face feature information into a preset three-dimensional conversion neural network model for three-dimensional conversion processing, so as to obtain the user's primary face model;

The effect rendering module is configured to obtain preset rendering data, and perform effect rendering on the primary face model based on the preset rendering data, so as to obtain a 3D face model of the user.

In addition, in order to achieve the above object, the present invention also proposes a 3D human face model generation device, which includes: a memory, a processor, and a 3D human face model stored in the memory and operable on the processor A generation program, the 3D face model generation program is configured to implement the steps of the method for generating a 3D face model as described above.

In addition, in order to achieve the above object, the present invention also proposes a storage medium, on which a 3D face model generation program is stored, and when the 3D face model generation program is executed by a processor, the 3D face model generation program as described above is realized. The steps of the face model generation method.

In the present invention, the face image information of the user is obtained, and the feature recognition is performed on the face image information to obtain the user's face feature information; the face feature information is input into a preset three-dimensional conversion neural network Perform three-dimensional conversion processing in the model to obtain the user's primary face model; obtain preset rendering data, and perform effect rendering on the primary face model based on the preset rendering data to obtain the user's 3D face model, compared with the prior art that directly models according to the user's face image information, the present invention uses the face feature information obtained based on the user's face image information in the preset 3D conversion neural network model Face model matching, and then adjust the matched face model to improve modeling efficiency, and then perform effect rendering on the obtained primary face model based on the preset rendering data, so as to improve the accuracy obtained based on the primary face model. The display effect of the 3D face model realizes the personalized setting of the 3D face model, and also improves the user's online social experience and increases user stickiness.

Description of drawings

Fig. 1 is the schematic structural diagram of the three-dimensional human face model generating equipment of the hardware operating environment that scheme of the embodiment of the present invention relates to;

Fig. 2 is a schematic flow chart of the first embodiment of the method for generating a three-dimensional face model of the present invention;

Fig. 3 is a schematic flow chart of the second embodiment of the method for generating a three-dimensional face model of the present invention;

Fig. 4 is a structural block diagram of the first embodiment of the device for generating a three-dimensional face model according to the present invention.

The realization of the purpose of the present invention, functional characteristics and advantages will be further described in conjunction with the embodiments and with reference to the accompanying drawings.

Detailed ways

It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Referring to FIG. 1 , FIG. 1 is a schematic structural diagram of a 3D face model generation device in a hardware operating environment involved in an embodiment of the present invention.

As shown in Figure 1, the 3D face model generating device may include: a processor 1001, such as a central processing unit (Central Processing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein, the communication bus 1002 is used to realize connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001 .

Those skilled in the art can understand that the structure shown in Figure 1 does not constitute a limitation to the three-dimensional face model generation device, and may include more or less components than those shown in the illustration, or combine some components, or different components layout.

As shown in FIG. 1 , the memory 1005 as a storage medium may include an operating system, a data storage module, a network communication module, a user interface module, and a 3D face model generation program.

In the three-dimensional face model generation device shown in Figure 1, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processing in the three-dimensional face model generation device of the present invention The processor 1001 and the memory 1005 can be set in the 3D face model generation device, and the 3D face model generation device calls the 3D face model generation program stored in the memory 1005 through the processor 1001, and executes the 3D face model generation program provided by the embodiment of the present invention. Face model generation method.

An embodiment of the present invention provides a method for generating a three-dimensional face model. Referring to FIG. 2 , FIG. 2 is a schematic flowchart of a first embodiment of the method for generating a three-dimensional face model according to the present invention.

In this embodiment, the method for generating a three-dimensional human face model includes the following steps:

Step S10: Obtain the user's face image information, and perform feature recognition on the face image information to obtain the user's face feature information;

It is easy to understand that, in this embodiment, the face image information can be obtained based on the user's self-portrait, or directly enable the built-in or external camera device of the above-mentioned 3D face model generation device to scan the image of the user, Obtain the user's face image information, and then perform image recognition on the face image information based on a preset face recognition algorithm to obtain key points of the face image, and the preset face recognition algorithm can be set according to actual needs , in this embodiment, in order to improve the efficiency of face recognition, to further improve the generation efficiency of the three-dimensional face model, it can be set to a template-based face recognition algorithm, such as, based on principal component analysis (Principal Component Analysis, PCA) Face recognition algorithm, that is, Eigenface algorithm. The key points of the human face image can be understood as feature points at key positions of the human face image, such as the contours of facial features (eyebrows, ears, nose, mouth), facial edge contours, beards, stains, moles, etc., which are not discussed in this embodiment. be restricted.

Further, the key points of the human face image can be transformed into the preset human face coordinate system, so as to obtain the key points between the key points of the human face image and the corresponding preset image key points in the preset human face coordinate system. point coordinate difference, and then generate the face feature information of the user based on the key point coordinate difference, the preset face coordinate system is a coordinate system based on a preset standard model, and the preset standard model can be It is understood as a face model standard built into the preset three-dimensional conversion neural network model, such as a face model based on a Marquardt mask, a face model based on an average face, etc. Without limitation, correspondingly, the key points of the preset image can be understood as feature points at the key positions of the preset standard model, such as facial features outlines, face outlines, etc., which are not limited in this embodiment. The preset 3D conversion neural network model can be understood as a neural network model established based on Convolutional Neural Networks (CNN) to perform face model matching and face model adjustment based on face feature information.

Step S20: Input the face feature information into the preset three-dimensional transformation neural network model to perform three-dimensional transformation processing, so as to obtain the user's primary face model;

It is easy to understand that the so-called face feature information is based on the key points between the key points of the face image corresponding to the obtained face image information and the key points of the preset image corresponding to the preset standard model in the preset face coordinate system The user's face feature information generated by the coordinate difference can be understood as the coordinate value corresponding to the key point of the face image generated based on the coordinate value of the key point of the preset image and the coordinate difference of the key point, which is used to represent the key point of the preset image. The gap between the points, that is, the feature of the user's face image information compared with the preset standard model, for example, the coordinate value corresponding to a preset key point of the preset standard model is (a, b), this The coordinate difference of the key point corresponding to the preset key point is (-0.90, +1.21), and the coordinate value of the face feature information corresponding to the user's face image information is (a-0.90, b+1.21).

It should be understood that the above is only an example, and does not constitute any limitation to the technical solution of the present invention. In specific applications, those skilled in the art can make settings according to needs, and the present invention is not limited thereto.

Further, after obtaining the facial feature information, in order to improve modeling efficiency, the facial feature information can be input into a preset 3D conversion neural network model for 3D conversion processing, so as to obtain the user's primary Face model, the so-called three-dimensional conversion processing, can be understood as performing face model matching in the preset three-dimensional conversion neural network model based on face feature information, and then performing model adjustment on the matched face model to obtain the user's Basic face model.

Step S30: Obtain preset rendering data, and perform effect rendering on the primary face model based on the preset rendering data, so as to obtain a 3D face model of the user.

It should be noted that the preset rendering data includes preset texture data and preset style data, and the preset texture data can be understood as the data used to render the facial epidermis and soft tissue structure of the primary face model, Such as skin texture, skin chroma, model plumpness, etc., the preset style data can be understood as the data used to render the style features of the primary face model, such as hue, exposure, contrast, etc., in specific implementation , in order to improve the display effect of the obtained 3D face model, the preset texture data and the preset style data corresponding to each face area can be obtained, as well as the edge point information and normal vector information corresponding to each face area, Wherein, the edge point information can be used to control the rendering range, the normal vector information can be used to control the rendering level, the combination of the edge point information and the normal vector information can be used to control the rendering intensity, and then based on the corresponding The preset texture data, the edge point information, and the normal vector information map the primary face model to obtain a face texture model, and then based on the preset style data corresponding to each face area, The edge point information and the normal vector information render the face texture model to obtain a three-dimensional face model of the user.

In a specific implementation, in order to further improve the display effect of the three-dimensional face model, to improve the user's network social experience and increase user viscosity, one type of preference information in the adjustment preference information input by the user can be obtained, and based on the one type of preference information information to adjust the primary face model to obtain a fine-tuned model for users. The type of preference information can be understood as the user’s preference data when adjusting the face contour, facial features contour, etc., such as the degree of face thinning , binocular magnification degree, nose reduction degree, etc. Then, the preset rendering data and the second-type preference information in the adjustment preference information can also be obtained, and the second-type preference information can be understood as the preference data when adjusting the skin texture , such as the degree of whitening/tanning, the degree of rosy cheeks, etc., and perform effect rendering on the fine-tuned model of the user based on the rendering data and the second-type preference information, so as to obtain the three-dimensional face model of the user.

It should be understood that the above is only an example, and does not constitute any limitation to the technical solution of the present invention. In specific applications, those skilled in the art can make settings according to needs, and the present invention is not limited thereto.

In this embodiment, the face image information of the user is obtained, and the feature recognition is performed on the face image information to obtain the user's face feature information; the face feature information is input to the preset three-dimensional conversion neuron Perform three-dimensional conversion processing in the network model to obtain the user's primary face model; obtain preset rendering data, and perform effect rendering on the primary face model based on the preset rendering data to obtain the The user's 3D face model is compared with the existing technology that directly models the user's face image information. In this example, the face feature information obtained based on the user's face image information is converted into the preset 3D neural network model. Perform face model matching, and then adjust the matched face model to improve modeling efficiency, and then perform effect rendering on the obtained primary face model based on the preset rendering data to improve the accuracy obtained based on the primary face model. The display effect of the 3D face model realizes the personalized setting of the 3D face model, and also improves the user's online social experience and increases user stickiness.

Referring to FIG. 3 , FIG. 3 is a schematic flowchart of a second embodiment of a method for generating a 3D face model according to the present invention.

Based on the first embodiment above, in this embodiment, the step S20 includes:

Step S201: Perform model matching between the face feature information and the model feature information of each preset face model in the preset three-dimensional transformation neural network model, and obtain a model between the face feature information and the model feature information suitability;

It should be noted that after obtaining the face feature information generated based on the user's face image information, the face feature information can be compared with the model feature information of each preset face model in the preset three-dimensional transformation neural network model. Model matching, to obtain the model matching degree between the facial feature information and the model feature information, the preset three-dimensional conversion neural network model can be understood as based on the convolutional neural network to perform based on the facial feature information A neural network model for face model matching and face model adjustment, the so-called face feature information, that is, the key points of the face image based on the obtained face image information correspond to the preset standard model in the preset face coordinate system The user's face feature information generated by the key point coordinate difference between the preset image key points can be understood as the corresponding key point of the face image generated based on the coordinate value of the preset image key point and the key point coordinate difference The coordinate value is used to indicate the gap between the key points of the preset image, that is, the feature of the user's face image information compared with the preset standard model, for example, a preset key point of the preset standard model corresponds to The coordinate value of the key point is (m, n), and the key point coordinate difference value corresponding to this preset key point is (-0.07, +0.23), then the coordinate value of the face feature information corresponding to the user's face image information is (m -0.07, n+0.23). The preset standard model can be understood as a face model standard built into the preset three-dimensional transformation neural network model, such as a face model based on a Marquardt mask, a face model based on an average face, etc. The embodiment does not limit this. Correspondingly, the key points of the preset image can be understood as the feature points at the key positions of the preset standard model, such as the outline of facial features, the outline of the face, etc., which are not included in this embodiment. limit.

It is easy to understand that each preset face model can also be understood as a face model set based on a preset standard model, and correspondingly, the model feature information of each preset face model can be understood as being used to represent the preset face model The characteristic information of the different gap segments between the key points of the model image of the preset standard model and the preset image key points of the preset standard model. Different face models are set in the fluctuation intervals, that is, the different fluctuation intervals of the coordinate values of the preset key points of the preset standard model are combined according to the preset combination method to obtain different face models. The combination method can be based on big data analysis of the face images stored in the preset face image database, and then based on the combination method in which the proportion in the big data analysis result is greater than the preset proportion, the preset face image database It is a database for storing face images that is updated in real time. Correspondingly, the preset combination method can also be periodically updated with the update of the preset face image database, and the preset ratio can be based on actual needs. Set, which is not limited in this embodiment.

Step S202: sort the model matching degree to obtain a matching degree ranking result, and determine a target face model based on the matching degree ranking result;

Further, in order to improve modeling efficiency, after obtaining the model matching degree between the face feature information and the model feature information of each preset face model in the preset three-dimensional conversion neural network model, the model can be matched Degrees are sorted from large to small to obtain matching degree sorting results, and it is judged whether the model matching degree corresponding to the preset face model with the first ranking order in the matching degree sorting results is greater than the preset matching degree, and the predetermined matching degree The matching degree can be set according to actual needs, such as 50%, 70%, etc., which is not limited in this embodiment. When the matching degree is greater than or equal to the preset matching degree, the preset person who ranks first in the ranking The face model is used as the target face model, and when the matching degree is less than the preset standard model, the preset standard model is used as the target face model.

Step S203: Adaptively adjust the target face model based on the face feature information to obtain a primary face model of the user.

It is easy to understand that after the target face model is obtained, the face feature information corresponding to the face feature value in each face area can be obtained, and the target face model corresponds to the face target in each preset partition. Value, the face feature value, can be understood as the coordinate value of the key points of the face image distributed in each face area, and correspondingly, the target face value can be understood as the target face model corresponding to each preset Set the coordinate values of the key points of the model image of the partition, compare the feature value of the human face with the target value of the human face, and obtain the feature comparison result, that is, obtain the key points of the human face image and the model image of each human face area The coordinate difference between the key points, and then, based on the feature comparison result, each face area of the target face model is adaptively adjusted to obtain the user's primary face model, that is, the The coordinate values of the key points of each model image are adjusted according to the coordinate values of the key points of the face image according to a preset adjustment ratio, and the preset ratio can be set according to actual needs, such as 75%, 85%, and 95%. In the specific implementation, because the generated primary face model and the subsequent 3D face model generated based on the primary face model are essentially non-selfie (that is, not WYSIWYG) 3D digital models, As the degree of anthropomorphism increases, the favorability of human beings does not always increase. They will show a sense of fear when they are very close but not completely consistent, and what the user himself hopes to show may not be completely true to himself, but to be able to represent himself. Characteristic three-dimensional digital images, even beautified three-dimensional digital images, therefore, in this embodiment, the preset ratio will not be set to 100%, that is, the coordinate values of the key points of each model image will be directly adjusted to the The coordinate values of the key points of the face image, because the Marquardt mask on which the preset standard model is based, or the average face, it is recognized that the closer the model standard is, the more beautiful it will be. Therefore, in order to improve the display effect, realize The personalized setting of the three-dimensional face model improves the user's network social experience and increases user viscosity. In this embodiment, after the target face model is obtained, the coordinate values of the key points of each model image of the target face model Adjusting according to the preset adjustment ratio according to the coordinate values of the key points of the human face image to obtain a primary human face model.

In this embodiment, the face feature information is model-matched with the model feature information of each preset face model in the preset three-dimensional transformation neural network model, and the relationship between the face feature information and the model feature information is obtained. the model matching degree; sort the model matching degree to obtain the matching degree sorting result, and determine the target face model based on the matching degree sorting result; perform the target human face model based on the facial feature information Adaptive adjustment to obtain the user's primary face model. Adaptively adjust each face area of the target face model based on the coordinate difference between the key points of the face image of each face area and the key points of the model image, so as to obtain the primary face model of the user, That is, the coordinate values of the key points of each model image are adjusted according to the coordinate values of the key points of the human face image according to a preset adjustment ratio, so as to improve the aesthetics of the subsequent three-dimensional human face model obtained based on the primary human face model, and also The personalized setting of the 3D face model is realized, which improves the user's online social experience and increases user stickiness.

In addition, an embodiment of the present invention also proposes a storage medium, on which a 3D face model generation program is stored, and when the 3D face model generation program is executed by a processor, the 3D face model as described above is realized The steps to generate the method.

Referring to FIG. 4 , FIG. 4 is a structural block diagram of the first embodiment of the device for generating a three-dimensional face model according to the present invention.

As shown in Figure 4, the three-dimensional face model generating device proposed by the embodiment of the present invention includes:

The feature recognition module 10 is used to obtain the face image information of the user, and carry out feature recognition to the face image information to obtain the face feature information of the user;

A three-dimensional conversion module 20, configured to input the face feature information into a preset three-dimensional conversion neural network model for three-dimensional conversion processing, so as to obtain the user's primary face model;

The effect rendering module 30 is configured to obtain preset rendering data, and perform effect rendering on the primary human face model based on the preset rendering data, so as to obtain a three-dimensional human face model of the user.

In this embodiment, the face image information of the user is obtained, and the feature recognition is performed on the face image information to obtain the user's face feature information; the face feature information is input to the preset three-dimensional conversion neuron Perform three-dimensional conversion processing in the network model to obtain the user's primary face model; obtain preset rendering data, and perform effect rendering on the primary face model based on the preset rendering data to obtain the The user's 3D face model is compared with the existing technology that directly models the user's face image information. In this example, the face feature information obtained based on the user's face image information is converted into the preset 3D neural network model. Perform face model matching, and then adjust the matched face model to improve modeling efficiency, and then perform effect rendering on the obtained primary face model based on the preset rendering data to improve the accuracy obtained based on the primary face model. The display effect of the 3D face model realizes the personalized setting of the 3D face model, and also improves the user's online social experience and increases user stickiness.

Based on the first embodiment of the above-mentioned 3D face model generating device of the present invention, a second embodiment of the 3D face model generating device of the present invention is proposed.

In this embodiment, the feature recognition module 10 is also used to obtain the user's face image information, and perform image recognition on the face image information to obtain key points of the face image;

The feature recognition module 10 is further configured to convert the key points of the human face image into a preset human face coordinate system, so as to obtain the corresponding preset value of the key points of the human face image and the preset human face coordinate system. Set the key point coordinate difference of the image key point;

The feature recognition module 10 is further configured to generate facial feature information of the user based on the key point coordinate difference.

The three-dimensional conversion module 20 is also used to perform model matching on the face feature information and the model feature information of each preset face model in the preset three-dimensional conversion neural network model, to obtain the face feature information and the The model matching degree between model feature information;

The three-dimensional conversion module 20 is further configured to sort the model matching degree to obtain a matching degree sorting result, and determine a target face model based on the matching degree sorting result;

The three-dimensional conversion module 20 is further configured to adaptively adjust the target face model based on the face feature information, so as to obtain the user's primary face model.

The three-dimensional conversion module 20 is also used to sort the matching degree of the models from large to small, so as to obtain the matching degree sorting result;

The three-dimensional conversion module 20 is also used to judge whether the model matching degree corresponding to the preset face model ranked first in the matching degree sorting result is greater than the preset matching degree;

The three-dimensional conversion module 20 is further configured to use the preset face model ranked first as the target face model when the preset matching degree is greater than or equal to the preset matching degree;

The three-dimensional conversion module 20 is further configured to use a preset standard model as the target face model when the matching degree is less than the preset matching degree.

The three-dimensional conversion module 20 is also used to obtain the face feature value corresponding to each face region of the face feature information, and the face target value corresponding to each preset partition of the target face model;

The three-dimensional conversion module 20 is further configured to compare the face feature value with the face target value to obtain a feature comparison result;

The three-dimensional conversion module 20 is further configured to adaptively adjust each face area of the target face model based on the feature comparison result, so as to obtain a primary face model of the user.

Wherein, the preset rendering data includes preset texture data and preset style data;

The effect rendering module 30 is further configured to acquire the preset texture data and the preset style data corresponding to each face area, as well as edge point information and normal vector information corresponding to each face area;

The effect rendering module 30 is further configured to map the primary face model based on the preset texture data corresponding to each face area, the edge point information and the normal vector information, to obtain a face texture Model;

The effect rendering module 30 is further configured to render the face texture model based on the preset style data corresponding to each face area, the edge point information and the normal vector information, and obtain the user's 3D face model.

The effect rendering module 30 is further configured to obtain one type of preference information in the adjustment preference information input by the user, and adjust the primary face model based on the one type of preference information to obtain a fine-tuned model by the user;

The effect rendering module 30 is further configured to obtain preset rendering data and second-type preference information in the adjustment preference information, and perform effect rendering on the user fine-tuning model based on the rendering data and the second-type preference information to obtain the 3D face model of the user.

For other embodiments or specific implementations of the device for generating a three-dimensional face model according to the present invention, reference may be made to the above-mentioned method embodiments, which will not be repeated here.

It should be noted that, as used herein, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article or system comprising a set of elements includes not only those elements, It also includes other elements not expressly listed, or elements inherent in the process, method, article, or system. Without further limitations, an element defined by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article or system comprising that element.

The serial numbers of the above embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general-purpose hardware platform, and of course also by hardware, but in many cases the former is better implementation. Based on this understanding, the technical solution of the present invention can be embodied in the form of software products in essence or in other words, the part that contributes to the prior art, and the computer software products are stored in a storage medium (such as read-only memory/random access memory, magnetic disk, optical disk), including several instructions to make a terminal device (which can be a mobile phone, computer, server, air conditioner, or network equipment, etc.) execute the methods described in various embodiments of the present invention.

The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process conversion made by using the description of the present invention and the contents of the accompanying drawings, or directly or indirectly used in other related technical fields , are all included in the scope of patent protection of the present invention in the same way.

Claims (9)

1. A three-dimensional face model generation method is characterized by comprising the following steps:

acquiring face image information of a user, and performing feature recognition on the face image information to acquire face feature information of the user;

inputting the face feature information into a preset three-dimensional conversion neural network model for three-dimensional conversion processing to obtain an initial-order face model of the user;

acquiring preset rendering data, and performing effect rendering on the initial-order face model based on the preset rendering data to obtain a three-dimensional face model of the user;

the step of obtaining the face image information of the user and performing feature recognition on the face image information to obtain the face feature information of the user specifically includes:

acquiring face image information of a user, and performing image recognition on the face image information to obtain face image key points;

converting the key points of the face image into a preset face coordinate system to obtain key point coordinate difference values of the key points of the face image and corresponding preset image key points in the preset face coordinate system;

and generating face feature information of the user based on the coordinate difference value of the key point.

2. The method for generating a three-dimensional face model according to claim 1, wherein the step of inputting the face feature information into a preset three-dimensional transformation neural network model for three-dimensional transformation processing to obtain an initial face model of the user specifically comprises:

performing model matching on the face feature information and model feature information of each preset face model in a preset three-dimensional conversion neural network model to obtain a model matching degree between the face feature information and the model feature information;

sorting the model matching degrees to obtain a matching degree sorting result, and determining a target face model based on the matching degree sorting result;

and carrying out self-adaptive adjustment on the target face model based on the face feature information to obtain an initial-order face model of the user.

3. The method for generating a three-dimensional face model according to claim 2, wherein the step of ranking the model matching degrees to obtain a matching degree ranking result and determining the target face model based on the matching degree ranking result specifically comprises:

sorting the model matching degrees from large to small to obtain a matching degree sorting result;

judging whether the model matching degree corresponding to the preset face model with the first sorting order in the matching degree sorting result is greater than the preset matching degree or not;

when the preset matching degree is greater than or equal to the preset matching degree, taking the preset face model with the first ordering order as a target face model;

and when the matching degree is smaller than the preset matching degree, taking a preset standard model as the target face model.

4. The method for generating a three-dimensional face model according to claim 2, wherein the step of performing adaptive adjustment on the target face model based on the face feature information to obtain an initial-order face model of the user specifically comprises:

acquiring face characteristic values of the face characteristic information corresponding to the face regions and face target values of the target face model corresponding to the preset subareas;

comparing the face characteristic value with the face target value to obtain a characteristic comparison result;

and performing self-adaptive adjustment on each face region of the target face model based on the feature comparison result to obtain an initial-order face model of the user.

5. The three-dimensional face model generation method according to any one of claims 1 to 4, wherein the preset rendering data includes preset texture data and preset style data;

correspondingly, the step of obtaining preset rendering data and performing effect rendering on the initial-order face model based on the preset rendering data to obtain the three-dimensional face model of the user specifically includes:

acquiring the preset texture data and the preset style data corresponding to each face area, and the edge point information and the normal vector information corresponding to each face area;

mapping the initial-order face model based on the preset texture data, the edge point information and the normal vector information corresponding to each face region to obtain a face texture model;

rendering the face texture model based on the preset style data, the edge point information and the normal vector information corresponding to each face area to obtain a three-dimensional face model of the user.

6. The method for generating a three-dimensional face model according to any one of claims 1 to 4, wherein the step of obtaining preset rendering data and performing effect rendering on the preliminary-stage face model based on the preset rendering data to obtain the three-dimensional face model of the user specifically includes:

acquiring one type of preference information in adjustment preference information input by a user, and adjusting the initial-order face model based on the one type of preference information to obtain a user fine-tuning model;

and acquiring preset rendering data and two types of preference information in the adjustment preference information, and performing effect rendering on the user fine-tuning model based on the rendering data and the two types of preference information to obtain the three-dimensional face model of the user.

7. A three-dimensional face model generation apparatus, characterized by comprising:

the characteristic identification module is used for acquiring face image information of a user and carrying out characteristic identification on the face image information to acquire face characteristic information of the user;

the three-dimensional conversion module is used for inputting the face feature information into a preset three-dimensional conversion neural network model for three-dimensional conversion processing so as to obtain an initial face model of the user;

the effect rendering module is used for acquiring preset rendering data and performing effect rendering on the initial-order face model based on the preset rendering data so as to obtain a three-dimensional face model of the user;

the feature recognition module is further configured to acquire face image information of a user, perform image recognition on the face image information, and acquire a face image key point; converting the key points of the face image into a preset face coordinate system to obtain key point coordinate difference values of the key points of the face image and corresponding preset image key points in the preset face coordinate system; and generating the face feature information of the user based on the coordinate difference value of the key point.

8. A three-dimensional face model generation apparatus, characterized in that the apparatus comprises: a memory, a processor and a three-dimensional face model generation program stored on the memory and executable on the processor, the three-dimensional face model generation program being configured to implement the steps of the three-dimensional face model generation method of any one of claims 1 to 6.

9. A storage medium, characterized in that a three-dimensional face model generation program is stored thereon, which when executed by a processor implements the steps of the three-dimensional face model generation method according to any one of claims 1 to 6.

Images