3D face similarity calculation and evaluation method based on electrostatic field and cloud model

Abstract

The present invention provides a method for calculating and evaluating three-dimensional facial similarity based on an electrostatic field and cloud model. The method comprises: obtaining a three-dimensional facial model and a similarity result calculated based on three-dimensional facial feature information; converting the three-dimensional facial model into a three-dimensional electrostatic field model; determining the point charge distribution of facial vertices and facial equipotential surfaces based on the three-dimensional electrostatic field model; performing computational processing on the point charge distribution of facial vertices and the facial equipotential surfaces to obtain a similarity result calculated based on the three-dimensional electrostatic field information; and establishing a cloud model based on the similarity result calculated based on the three-dimensional facial feature information and the similarity result calculated based on the three-dimensional electrostatic field information, wherein the cloud model is used to evaluate the accuracy of the similarity result. The method for calculating and evaluating three-dimensional facial similarity based on an electrostatic field and cloud model of the present invention improves the accuracy and stability of facial similarity calculation and evaluation.

Description

Technical Field

[0001] The present invention relates to the field of electronic information technology, and in particular to a three-dimensional face similarity calculation and evaluation method based on an electrostatic field and a cloud model. Background Art

[0002] Facial similarity estimation, a technique for verifying identity by acquiring facial features, has been a hot research topic in recent years in fields such as artificial intelligence, computer vision, and psychology. Compared to other biometric features used for identification (such as irises and fingerprints), the human face is unique, consistent, and highly non-replicable, providing a stable foundation for identification. Facial similarity is finding increasingly widespread applications in areas such as criminal investigation, intelligent transportation, access control, and internet services. Facial similarity estimation involves theories and methods from multiple disciplines, including computer vision and psychology, and places high demands on the comprehensiveness of researchers' knowledge base.

[0003] In the existing technology, facial similarity estimation methods are mainly divided into two categories: feature-based methods and deep learning-based methods. Feature-based methods mainly extract the feature vectors of the three-dimensional face model and then calculate the distance between the two feature vectors to determine the similarity. Deep learning-based methods use deep neural networks to learn the feature representation of the face model. However, the three-dimensional features proposed by the existing technology cannot effectively distinguish the information of different differences between faces. In addition, similarity estimation is a subjective and ambiguous task. Existing methods only provide a similarity measure for different faces, but there is no unified evaluation standard to assess the accuracy of each method.

[0004] Therefore, the accuracy and stability of face similarity estimation are poor. Summary of the Invention

[0005] In view of this, the present invention provides a three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model to solve the above problems.

[0006] According to a first aspect of the present invention, a method for calculating and evaluating three-dimensional face similarity based on an electrostatic field and a cloud model is provided, comprising: obtaining a three-dimensional face model and a similarity result calculated based on three-dimensional face feature information; converting the three-dimensional face model into a three-dimensional electrostatic field model; determining the point charge distribution of face vertices and the face equipotential surfaces based on the three-dimensional electrostatic field model; performing computational processing on the point charge distribution of face vertices and the face equipotential surfaces to obtain a similarity result calculated based on the three-dimensional electrostatic field information; and establishing a cloud model based on the similarity result calculated based on the three-dimensional face feature information and the similarity result calculated based on the three-dimensional electrostatic field information, wherein the cloud model is used to evaluate the accuracy of the similarity result.

[0007] In another implementation of the present invention, obtaining a similarity result calculated based on three-dimensional facial feature information includes: constructing geodesics on a three-dimensional facial model; determining feature points on the three-dimensional facial model based on the geodesics; and performing similarity calculation based on the feature points to obtain a similarity result calculated based on the three-dimensional facial feature information.

[0008] In another implementation of the present invention, a three-dimensional face model is converted into a three-dimensional electrostatic field model, including: representing the feature points on the three-dimensional face model as point charges to obtain a three-dimensional electrostatic field model; based on the three-dimensional electrostatic field model, determining the point charge distribution of the face vertices and the face equipotential surface, including: determining the point charge distribution of the face vertices based on the point charges in the three-dimensional electrostatic field model; performing calculations based on the vertex charge amount of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh; making the electric potential of each center point in the three-dimensional electrostatic field model equal to obtain a three-dimensional equipotential surface.

[0009] In another implementation of the present invention, the charge of each vertex of the face is expressed as:

[0010]

[0011] In another implementation of the present invention, a calculation is performed based on the vertex charge of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh, including: simplifying the electric potential of each triangular mesh in the three-dimensional electrostatic field model to the electric potential of the triangle center point, and the electric potential of the triangle center point is expressed as:

[0012] P(q′1,c j )+P(q′2,c j )+…+P(q′ n ,c j )=P cosnt

[0013] The electric potential of the center point of each triangular mesh is the sum of the electric potentials generated by all vertex charges.

[0014] In another implementation of the present invention, the three-dimensional equipotential surface is expressed as:

[0015]

[0016] Among them, P can be cosnt =1.

[0017] In another implementation of the present invention, the point charge distribution of the face vertices and the face equipotential surface are calculated and processed to obtain a similarity result based on the three-dimensional electrostatic field information, including: the three-dimensional face is represented by a charge descriptor as X = (q1, q2, ..., q n ); The similarity result calculated based on the three-dimensional electrostatic field information is calculated by the following formula:

[0018] Sim(X,X′)=1-JSD(X,X′)

[0019] where JSD(X,X′) represents the Jensen-Shannon divergence.

[0020] According to a second aspect of the present invention, a three-dimensional face similarity calculation and evaluation device based on an electrostatic field and a cloud model is provided, comprising: an acquisition module for acquiring a three-dimensional face model and a similarity result calculated based on three-dimensional face feature information; a processing module for converting the three-dimensional face model into a three-dimensional electrostatic field model; determining the point charge distribution of face vertices and the face equipotential surfaces based on the three-dimensional electrostatic field model; performing calculations on the point charge distribution of the vertices of the three-dimensional face model and the face equipotential surfaces to obtain a similarity result calculated based on the three-dimensional electrostatic field information; and an evaluation module for establishing a cloud model based on the similarity result calculated based on the three-dimensional face feature information and the similarity result calculated based on the three-dimensional electrostatic field information, wherein the cloud model is used to evaluate the accuracy of the similarity result.

[0021] According to a third aspect of the present invention, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for calculating and evaluating three-dimensional face similarity based on an electrostatic field and a cloud model as described above are implemented.

[0022] According to a fourth aspect of the present invention, a computer storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model are implemented as described in any of the above items.

[0023] In the three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model of the present invention, by utilizing the characteristics of electrostatic field, equipotential surfaces of faces are simulated to provide a similarity measure between different faces. The face similarity obtained by the three-dimensional electrostatic field model has a higher accuracy. Based on the cloud model principle, a unified standard for similarity estimation is given by integrating multiple different methods, which can be used to evaluate the accuracy of different methods, thereby improving the accuracy and stability of face similarity calculation and evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. By reading the detailed description of the embodiments below, the advantages and benefits of the solutions will become clear to those skilled in the art. The drawings are only for the purpose of illustrating preferred embodiments and are not to be considered as limiting the present invention. In the drawings:

[0025] Figure 1 The present invention is a flowchart of a method for calculating and evaluating three-dimensional face similarity based on an electrostatic field and a cloud model according to an embodiment of the present invention.

[0026] Figure 2 This is a flowchart of the steps for determining unified evaluation standards according to another embodiment of the present invention.

[0027] Figure 3 FIG. 1 is a schematic diagram of facial feature points according to another embodiment of the present invention.

[0028] Figure 4 FIG2 is a schematic diagram of the charge distribution of facial vertices and facial equipotential surfaces according to another embodiment of the present invention.

[0029] Figure 5 This is a schematic diagram of the cloud model principle of another embodiment of the present invention.

[0030] Figure 6 Schematic diagram of a similarity cloud model integrating four methods according to another embodiment of the present invention.

[0031] Figure 7 Schematic diagram of a cloud model generated based on each method according to another embodiment of the present invention.

[0032] Figure 8 FIG. 1 is a schematic diagram of grouping faces with obvious differences according to another embodiment of the present invention.

[0033] Figure 9 This is a structural block diagram of a three-dimensional face similarity calculation and evaluation device based on electrostatic field and cloud model according to another embodiment of the present invention.

[0034] Figure 10FIG. 4 is a schematic structural diagram of an electronic device according to another embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and detailedly described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments in the embodiments of the present invention should fall within the scope of protection of the embodiments of the present invention.

[0036] Figure 1 A flowchart of a method for calculating and evaluating three-dimensional face similarity based on an electrostatic field and cloud model is provided in an embodiment of the present invention, such as Figure 1 As shown, this embodiment mainly includes the following steps:

[0037] S101: Obtain a three-dimensional face model and a similarity result calculated based on three-dimensional face feature information.

[0038] For example, a 3D face model can be obtained through channels such as video data and facial recognition data. Existing face similarity estimation methods are used to calculate the similarity of two 3D face models, resulting in a similarity result based on the 3D facial feature information. Existing face similarity estimation methods include manual judgment, feature-based face similarity estimation, and deep learning-based face similarity estimation. This application does not specifically limit the existing face similarity estimation methods used.

[0039] S102: Convert the three-dimensional face model into a three-dimensional electrostatic field model.

[0040] For example, according to the existing method, each feature point on the three-dimensional face model is determined, such as Figure 3 The dots shown in the figure assume that each feature point is a point charge, where the charge of each feature point is q i , and obtain the three-dimensional electrostatic field model.

[0041] S103. Determine the point charge distribution of the face vertices and the face equipotential surfaces based on the three-dimensional electrostatic field model.

[0042] For example, if the charge of each point charge is distributed to the surrounding common vertices according to Gaussian distribution, the charge of each vertex of the face can be expressed as:

[0043]

[0044] For each triangular mesh of the face, its electric potential can be simplified to the electric potential of the triangle center point, then the center point cj The electric potential can be expressed as:

[0045] P(q′1,c j )+P(q′2,c j )+…+P(q′ n ,c j )=P cosnt

[0046] For the entire face, the electric potential of each triangle mesh center point is the sum of the electric potentials generated by all vertex charges, so that the electric potentials of all center points of the face are equal (i.e., P cosnt ), then the face can be simulated as a three-dimensional equipotential surface.

[0047] S104: Calculate and process the point charge distribution of the face vertices and the face equipotential surfaces to obtain a similarity result calculated based on the three-dimensional electrostatic field information.

[0048] For example, after solving the equipotential surface of each face, a three-dimensional face can be represented by a charge descriptor as X = (q1, q2, ..., q n ), so the similarity of two faces can be expressed by the formula:

[0049] Sim(X,X′)=1-JSD(X,X′)

[0050] where JSD(X,X′) represents the Jensen-Shannon divergence.

[0051] S105 , establishing a cloud model based on the similarity results calculated based on the three-dimensional facial feature information and the similarity results calculated based on the three-dimensional electrostatic field information, wherein the cloud model is used to evaluate the accuracy of the similarity results.

[0052] For example, Figure 2 As shown in the figure, a mapping relationship between qualitative concepts and quantitative similarities is established based on the cloud model, and a similarity evaluation index is established in a statistical way by combining multiple methods. This index can be used to compare different methods in similarity measurement.

[0053] In the three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model of the present invention, by utilizing the characteristics of electrostatic field, equipotential surfaces of faces are simulated to provide a similarity measure between different faces. The face similarity obtained by the three-dimensional electrostatic field model has a higher accuracy. Based on the cloud model principle, a unified standard for similarity estimation is given by integrating multiple different methods, which can be used to evaluate the accuracy of different methods, thereby improving the accuracy and stability of face similarity calculation and evaluation.

[0054] In another implementation of the present invention, obtaining a similarity result calculated based on three-dimensional facial feature information includes: constructing geodesics on a three-dimensional facial model; determining feature points on the three-dimensional facial model based on the geodesics; and performing similarity calculation based on the feature points to obtain a similarity result calculated based on the three-dimensional facial feature information.

[0055] For example, the similarity result calculated based on three-dimensional facial feature information can be calculated using the methods mentioned in the following documents. The methods mentioned in the following documents are for example only, and this application does not impose any specific restrictions on the methods used to obtain the similarity result calculated based on three-dimensional facial feature information.

[0056] Reference [1] (Zhao J, Liu C, Wu Z, et al. 3D facial similarity measure based on geodesic network and curvatures [J]. Mathematical Problems in Engineering, 2014, 2014) proposed a method based on a combination of geodesic network and curvature features. First, the nose tip position is determined, then geodesics are constructed along multiple angles, and finally four features are calculated, namely mean curvature, Gaussian curvature, shape index, and curvature, as similarity measures for each face.

[0057] Reference [2] (Hu Xiaojing, Zhou Mingquan, Geng Guohua, et al. Three-dimensional facial similarity measurement based on adaptive neighborhood descriptor [J]. Computer Engineering and Applications, 2019, 55(2): 187-192.) proposed a method for three-dimensional facial similarity measurement based on adaptive neighborhood feature descriptor. This method extracts facial feature points based on geodesics and enhances the feature description of feature points through adaptive neighborhood based on the multi-scale idea. Subsequently, the four geometric invariants of each feature point are calculated based on the neighborhood feature points, and the covariance is calculated as the final similarity measurement result.

[0058] Reference [3] (Bahri M, O'Sullivan E, Gong S, et al. Shape my face: registering 3D face scans by surface-to-surface translation [J]. International Journal of Computer Vision, 2021, 129(9): 2680-2713.) uses multi-view depth perception representation for 3D face similarity estimation, which can accurately depict facial meshes in multiple coordinates. In addition, they also proposed a training strategy that incorporates view specificity and region consistency to improve the reliability of the network when processing various projections.

[0059] The face similarity estimation method proposed in the literature [4] (Zhao JL, Wu ZK, Pan ZK, et al. 3D face similarity measure by fré chet distances of geodesics [J]. Journal of Computer Science and Technology, 2018, 33: 207-222.) is mainly divided into the following three steps: the geodesics on each 3D facial model emitted from the nose tip are extracted with the same initial direction and equal angle increments; the Fréchet distance between two sets of corresponding geodesics on the two facial models is calculated; the similarity between the two facial models is calculated based on the Fréchet distance of the geodesics obtained in the second step.

[0060] In another implementation of the present invention, a three-dimensional face model is converted into a three-dimensional electrostatic field model, including: representing the feature points on the three-dimensional face model as point charges to obtain a three-dimensional electrostatic field model; based on the three-dimensional electrostatic field model, determining the point charge distribution of the face vertices and the face equipotential surface, including: determining the point charge distribution of the face vertices based on the point charges in the three-dimensional electrostatic field model; performing calculations based on the vertex charge amount of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh; making the electric potential of each center point in the three-dimensional electrostatic field model equal to obtain a three-dimensional equipotential surface.

[0061] For example, the three-dimensional face model is converted into a three-dimensional electrostatic field model, and the concept of electric field in physics is introduced. By constructing an electrostatic field in space to express the face, it is used to estimate the similarity of the face. In real life, the special substance in the space around a particle with a fixed charge is usually called an electric field, and the electrostatic field is the electric field excited by a stationary charge (a charge that is stationary relative to the observer). The electrostatic field excited by a point charge with a charge q can add physical properties to a surrounding point p or a triangular facet ΔABC. We take the electric potential P(q,p) as the physical property of point p, then the electric potential can be expressed by the formula P(q,p) = kq / r, where k represents the electrostatic constant and r represents the Euclidean distance from the point p to the point charge.

[0062] It should be understood that according to the superposition principle, at multiple point charges {q1,q2,q3,…,q i ,…,q n}, the potential can be expressed as Psum(p)=∑P(q i ,p). Electrostatic fields have some unique properties in physics: first, the equipotential surfaces constructed by electrostatic fields are smooth and continuous, allowing us to construct different equipotential surfaces to fit different shapes by assigning different charge distributions to multiple point charges. Second, the electric field lines originating from the charges never intersect. This is because the direction of the electric field lines is consistent with the direction of the electric field vector. If two electric field lines intersect, it means that the electric field at the intersection has two different directions. However, Coulomb's law shows that the field strength at a point is unique, so electric field lines will never intersect.

[0063] In another implementation of the present invention, the charge of each vertex of the face is expressed as:

[0064]

[0065] In another implementation of the present invention, a calculation is performed based on the vertex charge of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh, including: simplifying the electric potential of each triangular mesh in the three-dimensional electrostatic field model to the electric potential of the triangle center point, and the electric potential of the triangle center point is expressed as:

[0066] P(q′1,c j )+P(q′2,c j )+…+P(q′ n ,c j )=P cosnt

[0067] The electric potential of the center point of each triangular mesh is the sum of the electric potentials generated by all vertex charges.

[0068] In another implementation of the present invention, the three-dimensional equipotential surface is expressed as:

[0069]

[0070] Among them, P can be cosnt =1.

[0071] For example, if the electric potential of all the center points of the face is equal (i.e., Pconst), the face can be simulated as a three-dimensional equipotential surface:

[0072]

[0073] Without loss of generality, we can let P cosnt =1, then solving the problem of face equipotential surface can be transformed into a problem of optimizing the minimum value of energy function, and the energy function is expressed as:

[0074]

[0075] In addition, a smoothing term is introduced:

[0076]

[0077] Among them, σ 2 (c j ) represents the center point c j The variance of the potential and its adjacent points, P j Indicates the potential value calculated from the marked point, P′ j Represents the electric potential value calculated after dispersing the charge using a Gaussian distribution.

[0078] The final objective function is expressed as:

[0079] Δ=E q′ +E s

[0080] While minimizing the objective function, it is necessary to constrain the charge of the point charge to ensure that it is greater than 0. Finally, the point charge distribution of the face vertices and the face equipotential surface can be estimated by optimization, such as Figure 4 shown.

[0081] In another implementation of the present invention, the point charge distribution of the face vertices and the face equipotential surface are calculated and processed to obtain a similarity result based on the three-dimensional electrostatic field information, including: the three-dimensional face is represented by the charge descriptor X = (q1, q2, ..., q n ); the similarity between two faces is calculated by the following formula:

[0082] Sim(X,X′)=1-JSD(X,X′)

[0083] where JSD(X,X′) represents the Jensen-Shannon divergence.

[0084] In another implementation of the present invention, Figure 5 As shown in the figure, we propose using a cloud model algorithm to comprehensively analyze the similarity results obtained by human judgment, the algorithm-derived similarity results, and qualitative results, and use this to evaluate the similarity algorithm. Compared with other methods, using the cloud model to construct cloud maps is more accurate and intuitive, while also providing a relatively objective reference standard in a statistical sense. Here, Ex is the expected value, En is the entropy, He is the excess entropy, FCG is the forward cloud generator, and BCG is the reverse cloud generator.

[0085] Using the cloud model principle, this method, the questionnaire results, and the methods of References 1 and 4 were combined to obtain Figure 6 The comprehensive cloud model shown in Figure 2 is used as a reference standard, and the cloud models obtained based on each method are used as reference standards. Figure 7 ) to compare and determine the similarity.

[0086] In another implementation of the present invention, feature information extracted based on the Euclidean distance of a 3D model can most directly reflect the surface characteristics of the model. The main purpose of the correlation analysis is to explore the relationship between the feature information extracted based on the electrostatic field and the Euclidean distance of the model, thereby demonstrating the feasibility of using this feature for 3D face similarity estimation. Table 1 shows the correlation results between our method and the Euclidean distance of the 3D model. The results show a moderate degree of correlation, thus being able to reflect 3D surface feature information to a certain extent and having the potential to describe shape characteristics.

[0087] Table 1 Experimental results of correlation with Euclidean distance

[0088] Correlation coefficient Spearman coefficient Pearson coefficient Electrostatic Field Method 0.43 0.40

[0089] In another implementation of the present invention, it is determined whether the extracted feature information can distinguish between two types of faces with obvious differences. Generally speaking, an excellent feature information has the ability to distinguish between faces with obvious differences. In this experiment, three groups of faces with obvious differences were selected from the BU-3DFE face dataset, each group containing a total of 7 neutral faces, such as Figure 8 shown.

[0090] Table 2 shows the t-test results. Generally, when the p-value is less than 0.05, we consider the null hypothesis to be rejected, indicating that there is a significant difference between the two groups. In all three groups, both the questionnaire results and this method can effectively distinguish 3D faces with significant differences.

[0091] Table 2. Results of the difference discrimination experiment

[0092]

[0093] In another implementation of the present invention, the similarity results from this method, the questionnaire, and the methods in References 1 and 4 were combined to generate a unified cloud model that can be used as an evaluation standard. The similarity between two cloud models can be measured using the Normal Cloud Model Similarity Calculation Method (MCM).

[0094] Table 3 shows the similarity between the cloud models generated by different methods and the unified standard. The larger the value, the closer the method is to the unified standard and the more accurate the similarity assessment is.

[0095] Table 3 MCM experimental results

[0096] method Very similar relatively similar Generally similar Dissimilar average This method 0.856 0.805 0.836 0.974 0.868 Questionnaire 0.622 0.685 0.660 0.633 0.650 Document 1 0.676 0.428 0.493 0.497 0.524 Document 4 0.579 0.589 0.773 0.913 0.714

[0097] The three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model of the present invention fits different faces through electrostatic field and can be used in fields such as face recognition, face comparison and medical imaging.

[0098] According to a second aspect of the present invention, there is provided a device 900 for calculating and evaluating three-dimensional face similarity based on an electrostatic field and a cloud model, characterized by comprising:

[0099] Acquisition module 901: used to obtain a three-dimensional face model and a similarity result calculated based on three-dimensional face feature information.

[0100] Processing module 902: Convert the three-dimensional face model into a three-dimensional electrostatic field model; based on the three-dimensional electrostatic field model, determine the point charge distribution of the face vertices and the face equipotential surfaces; calculate and process the point charge distribution of the three-dimensional face model vertices and the face equipotential surfaces to obtain a similarity result calculated based on the three-dimensional electrostatic field information.

[0101] Evaluation module 903: Building a cloud model based on the similarity results calculated based on the three-dimensional facial feature information and the similarity results calculated based on the three-dimensional electrostatic field information. The cloud model is used to evaluate the accuracy of the similarity results.

[0102] In the three-dimensional face similarity calculation and evaluation device based on electrostatic field and cloud model of the present invention, by utilizing the characteristics of electrostatic field, the facial equipotential surface is simulated to provide a similarity measurement between different faces. The face similarity obtained by the three-dimensional electrostatic field model has a higher accuracy. Based on the cloud model principle, a unified standard for similarity estimation is given by integrating multiple different methods, which can be used to evaluate the accuracy of different methods, thereby improving the accuracy and stability of face similarity calculation and evaluation.

[0103] In another implementation of the present invention, the acquisition module 901 is also used to construct geodesics on the three-dimensional face model; determine feature points on the three-dimensional face model based on the geodesics; and perform similarity calculation based on the feature points to obtain a similarity result calculated based on the three-dimensional face feature information.

[0104] In another implementation of the present invention, the processing module 902 is also used to represent the feature points on the three-dimensional face model as point charges to obtain a three-dimensional electrostatic field model; based on the point charges in the three-dimensional electrostatic field model, determine the point charge distribution of the face vertices; perform calculations based on the vertex charge of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh; make the electric potential of each center point in the three-dimensional electrostatic field model equal to obtain a three-dimensional equipotential surface.

[0105] In another implementation of the present invention, the processing module 902 is further configured to represent the charge amount of each vertex of the face:

[0106]

[0107] In another implementation of the present invention, the processing module 902 is further configured to simplify the electric potential of each triangular mesh in the three-dimensional electrostatic field model into the electric potential of the triangle center point. The electric potential of the triangular mesh center point is expressed as:

[0108] P(q′1,c j )+P(q′2,c j )+…+P(q′ n ,c j )=P cosnt

[0109] The electric potential of the center point of each triangular mesh is the sum of the electric potentials generated by all vertex charges.

[0110] In another implementation of the present invention, the processing module 902 is further configured to represent a three-dimensional equipotential surface:

[0111]

[0112] Among them, P can be cosnt =1.

[0113] In another implementation of the present invention, the processing module 902 is further configured to represent a three-dimensional face using a charge descriptor:

[0114] X=(q1,q2,…,q n )

[0115] The similarity between two faces is calculated using the following formula:

[0116] Sim(X,X′)=1-JSD(X,X′)

[0117] where JSD(X,X′) represents the Jensen-Shannon divergence.

[0118] like Figure 10 As shown, the electronic device 1000 may include: a processor (processor) 1001 , a memory (memory) 1003 , a communication bus 1004 , and a communication interface (Communications Interface) 1005 .

[0119] in:

[0120] The processor 1001 , the memory 1003 and the communication interface 1005 communicate with each other via the communication bus 1004 .

[0121] The communication interface 1005 is used to communicate with other electronic devices or servers.

[0122] Processor 1001 is used to execute program 1002, and specifically can execute the steps of any one of the three-dimensional face similarity calculation and evaluation methods based on electrostatic field and cloud model in the above embodiments.

[0123] Specifically, the program 1002 may include program codes, which include computer operating instructions.

[0124] The processor 1001 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the smart device may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0125] The memory 1003 is used to store the program 1002. The memory 1003 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0126] Program 1002 can be specifically used to enable the processor 1001 to execute to implement any of the steps of the three-dimensional face similarity calculation and evaluation method based on the electrostatic field and cloud model described in the embodiment. The specific implementation of each step in program 1002 can refer to the corresponding description of the steps and units executed by any of the three-dimensional face similarity calculation and evaluation methods based on the electrostatic field and cloud model in the above steps, which will not be repeated here. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described equipment and modules can refer to the corresponding process description in the aforementioned method embodiment.

[0127] The exemplary embodiments of the present application further provide a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the methods of the various embodiments of the present application.

[0128] The method according to the embodiment of the present invention described above can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk or magneto-optical disk), or as computer code that is originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded via a network and will be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor or programmable or dedicated hardware (such as an ASIC or FPGA). It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown here, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown here.

[0129] Thus far, specific embodiments of the present invention have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing may be advantageous.

[0130] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0131] In the description of the present invention, the terms "first" and "second" are used solely to facilitate description of different components or names and should not be construed as indicating or implying a sequential relationship, relative importance, or implicitly specifying the quantity of the technical features being described. Therefore, features specified as "first" or "second" may explicitly or implicitly include at least one of such features.

[0132] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0133] It should be noted that although the specific embodiments of the present invention are described in detail in conjunction with the accompanying drawings, this should not be construed as limiting the scope of protection of the present invention. Within the scope described by the claims, various modifications and variations that can be made by those skilled in the art without creative effort still fall within the scope of protection of the present invention.

[0134] The examples of the embodiments of the present invention are intended to briefly illustrate the technical features of the embodiments of the present invention so that those skilled in the art can intuitively understand the technical features of the embodiments of the present invention, and are not intended to improperly limit the embodiments of the present invention.

[0135] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model, characterized in that: include: Obtaining a three-dimensional face model and a similarity result calculated based on three-dimensional face feature information; Converting the three-dimensional face model into a three-dimensional electrostatic field model includes: Representing feature points on the three-dimensional face model as point charges to obtain a three-dimensional electrostatic field model; Based on the three-dimensional electrostatic field model, determining the point charge distribution of the face vertices and the face equipotential surface includes: determining the point charge distribution of the vertices of the face based on the point charges in the three-dimensional electrostatic field model; Calculating and processing the vertex charge of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh; Setting the electric potential of each center point in the three-dimensional electrostatic field model to be equal to obtain a three-dimensional equipotential surface; The charge of each vertex of the face is expressed as: Among them, q i Represents the charge of adjacent feature points, dis(q i ,q′) represents the Euler distance between two points, σ q Indicates the preset coefficient constant; Performing calculations based on the point charge distribution of the face vertices and the face equipotential surfaces to obtain a similarity result calculated based on three-dimensional electrostatic field information; A cloud model is established based on the similarity result calculated based on the three-dimensional facial feature information and the similarity result calculated based on the three-dimensional electrostatic field information. The cloud model is used to evaluate the accuracy of the similarity result.

2. The method according to claim 1, characterized in that The obtaining of a similarity result calculated based on the three-dimensional facial feature information includes: constructing geodesics on the three-dimensional face model; determining feature points on the three-dimensional face model based on geodesics; A similarity calculation is performed based on the feature points to obtain a similarity result calculated based on the three-dimensional facial feature information.

3. The method according to claim 2, characterized in that The calculating and processing of the vertex charge of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh includes: The electric potential of each triangular mesh in the three-dimensional electrostatic field model is simplified to the electric potential of the triangle center point, and the electric potential of the triangular mesh center point is expressed as: P(q′1,c j )+P(q′2,c j )+…+P(q ′ n ,c j )=P cosnt Among them, let P cosnt =1; P(q ′ n ,c j ) represents vertex q ′ n At the center of the triangle mesh c j The electric potential generated, the electric potential of each triangle mesh center point is the sum of the electric potentials generated by all vertex charges.

4. The method according to claim 3, characterized in that The three-dimensional equipotential surface is expressed as: Among them, P(q ′ n ,c m ) represents vertex q ′ n At the center of the triangle mesh c m The electric potential generated.

5. The method according to claim 1, wherein The calculation and processing based on the point charge distribution of the face vertices and the face equipotential surface to obtain a similarity result calculated based on the three-dimensional electrostatic field information includes: The three-dimensional face is represented by the charge descriptor X = (q1, q2, ..., q n ); The similarity result calculated based on the three-dimensional electrostatic field information is calculated by the following formula: Sim(X,X ′ )=1-JSD(X,X ′ ) Among them, JSD(X,X ′ ) denotes the Jensen-Shannon divergence.

6. A three-dimensional face similarity calculation and evaluation device based on electrostatic field and cloud model, characterized in that: include: Acquisition module: used to obtain the three-dimensional face model and the similarity result calculated based on the three-dimensional face feature information; Processing module: converting the three-dimensional face model into a three-dimensional electrostatic field model, including: representing the feature points on the three-dimensional face model as point charges to obtain a three-dimensional electrostatic field model; determining the point charge distribution of the face vertices and the face equipotential surface based on the three-dimensional electrostatic field model, including: determining the point charge distribution of the face vertices based on the point charges in the three-dimensional electrostatic field model; performing calculations based on the vertex charge amount of each triangular mesh in the three-dimensional electrostatic field model to obtain the electric potential of the center point of each triangular mesh; making the electric potential of each center point in the three-dimensional electrostatic field model equal to obtain a three-dimensional equipotential surface; the charge amount of each vertex of the face is expressed as: Among them, q i Represents the charge of adjacent feature points, dis(q i ,q′) represents the Euler distance between two points, σ q Indicates the preset coefficient constant; Performing calculations based on the point charge distribution of the face vertices and the face equipotential surfaces to obtain a similarity result calculated based on three-dimensional electrostatic field information; Evaluation module: establishing a cloud model according to the similarity result calculated based on the three-dimensional facial feature information and the similarity result calculated based on the three-dimensional electrostatic field information, wherein the cloud model is used to evaluate the accuracy of the similarity result.

7. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for calculating and evaluating three-dimensional face similarity based on an electrostatic field and a cloud model are implemented as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that The computer storage medium stores a computer program, which, when executed by a processor, implements the steps in the three-dimensional face similarity calculation and evaluation method based on electrostatic field and cloud model as described in any one of claims 1 to 5.

Images