Three-dimensional texture map grid quality evaluation method and related device

Abstract

The embodiment of the invention provides a three-dimensional texture map grid quality evaluation method and a related device, and relates to the fields of video coding and decoding, cloud technology, artificial intelligence and the like. The method comprises the steps of obtaining a to-be-evaluated distortion grid and a reference grid corresponding to the distortion grid, sampling points in at least one to-be-processed grid to obtain a target point set, for each target point in the target point set, based on the distance between the points, determining a first local geodesic line area corresponding to the target point from the reference grid, and determining a second local geodesic line area corresponding to the first local geodesic line area; determining a second local geodesic line area corresponding to the target point from the distortion grid; for each target point, determining the similarity between the first local geodesic line area and the second local geodesic line area corresponding to the target point; and determining a quality evaluation result of the distorted grid based on the similarity corresponding to each target point. According to the embodiment of the invention, the applicability of grid quality evaluation is realized.

Description

Technical Field

[0001] The present application relates to the field of video coding and decoding technology, and may also relate to the fields of cloud technology, artificial intelligence, etc. Specifically, the present application relates to a quality evaluation method and related device for a three-dimensional texture grid. Background Art

[0002] With the application of 3D mesh models in the field of video coding and decoding, such as game entertainment, film and television animation, virtual reality and other fields, there are more and more processing processes for 3D mesh models, including embedding watermarks, compressing meshes, simplifying meshes, etc. These processing processes will inevitably cause mesh distortion and affect people's subjective visual experience. The quality evaluation of 3D meshes plays a very important role in evaluating and optimizing the acquisition and processing algorithms of 3D models and reducing the distortion of 3D models. At present, the main method is the point-based objective evaluation method: this method mainly converts the 3D mesh into a 3D point cloud by sampling, and then uses the point cloud objective evaluation algorithm to calculate the objective score and use it as the 3D mesh objective score.

[0003] However, the current point cloud-based evaluation methods have requirements for the topological structure of the mesh, such as requiring the distorted and reference meshes to have the same point density, the same point connection relationship, and the same level of detail. Otherwise, the corresponding features cannot be extracted or the extracted feature values ​​have obvious errors, resulting in insufficient applicability. Summary of the invention

[0004] The embodiment of the present application provides a quality evaluation method and related device for a three-dimensional texture map mesh, which are used to solve the technical problem that the applicability of existing mesh quality evaluation is not high enough.

[0005] On the one hand, an embodiment of the present application provides a quality evaluation method for a three-dimensional texture grid, comprising:

[0006] Obtaining a distorted mesh to be evaluated and a reference mesh corresponding to the distorted mesh, where the reference mesh and the distorted mesh are three-dimensional meshes corresponding to the texture image;

[0007] Sampling points in at least one to-be-processed grid to obtain a target point set, wherein the at least one to-be-processed grid is at least one of a reference grid and a distorted grid;

[0008] For each target point in the target point set, based on the distance between the points, determine a first nearest point of the target point from the reference grid and a second nearest point of the target point from the distorted grid;

[0009] For each target point, the first nearest point of the target point is used as the center point of the region, and a first local geodesic region corresponding to the target point is determined from the reference grid; and the second nearest point of the target point is used as the center point of the region, and a second local geodesic region corresponding to the target point is determined from the distorted grid;

[0010] For each target point, determining the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point;

[0011] Based on the similarity corresponding to each target point, the quality evaluation result of the distorted grid is determined.

[0012] On the other hand, the embodiment of the present application further provides a quality evaluation device for a three-dimensional texture grid, comprising:

[0013] A mesh acquisition module is used to acquire a distorted mesh to be evaluated and a reference mesh corresponding to the distorted mesh, where the reference mesh and the distorted mesh are three-dimensional meshes corresponding to the texture image;

[0014] A sampling module, used for sampling points in at least one to-be-processed grid to obtain a target point set, wherein the at least one to-be-processed grid is at least one of a reference grid and a distorted grid;

[0015] A neighbor point determination module, for each target point in the target point set, determining a first nearest neighbor point of the target point from the reference grid based on the distance between the points, and determining a second nearest neighbor point of the target point from the distorted grid;

[0016] A local area determination module is used for, for each target point, taking the first nearest point of the target point as the area center point, determining a first local geodesic area corresponding to the target point from the reference grid, and taking the second nearest point of the target point as the area center point, determining a second local geodesic area corresponding to the target point from the distorted grid;

[0017] A similarity determination module, configured to determine, for each target point, a similarity between a first local geodesic region and a second local geodesic region corresponding to the target point;

[0018] The quality evaluation module is used to determine the quality evaluation result of the distorted grid based on the similarity corresponding to each target point.

[0019] On the other hand, an embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any aspect of the method.

[0020] On the other hand, an embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of any aspect of the method when the computer program is executed by a processor.

[0021] On the other hand, an embodiment of the present application further provides a computer program product, including a computer program, which implements the steps of any aspect of the method when the computer program is executed by a processor.

[0022] The beneficial effects of the technical solution provided by the embodiment of the present application are:

[0023] By obtaining a distorted grid to be evaluated and a reference grid corresponding to the distorted grid, the reference grid and the distorted grid are three-dimensional grids corresponding to the texture map; sampling points in at least one grid to be processed to obtain a target point set, the at least one grid to be processed is at least one of the reference grid and the distorted grid; for each target point in the target point set, based on the distance between the points, determining the first nearest point of the target point from the reference grid, and determining the second nearest point of the target point from the distorted grid; for each target point, taking the first nearest point of the target point as the center point of the region, determining the first local geodesic region corresponding to the target point from the reference grid, and taking the second nearest point of the target point as the center point of the region, determining the second local geodesic region corresponding to the target point from the distorted grid; for each target point, determining the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point; based on the similarity corresponding to each target point, determining the quality evaluation result of the distorted grid, the topological structure of the grid can be arbitrary, thereby improving the applicability of three-dimensional grid evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0025] Figure 1 A schematic diagram of a system architecture provided for an embodiment of the present application;

[0026] Figure 2 A schematic diagram of a process for evaluating the quality of a three-dimensional texture grid provided in an embodiment of the present application;

[0027] Figure 3 A schematic diagram of a reference grid or a distorted grid provided in an embodiment of the present application;

[0028] Figure 4is a schematic diagram of a local geodesic area determined according to one-step neighboring points provided in an embodiment of the present application;

[0029] Figure 5 A schematic diagram of a process for obtaining a smoothness feature provided in an embodiment of the present application;

[0030] Figure 6 A schematic diagram of an enclosed area corresponding to a patch in a texture map provided in an embodiment of the present application;

[0031] Figure 7 A schematic diagram of a process for obtaining texture map features provided in an embodiment of the present application;

[0032] Figure 8 A schematic diagram of a process for obtaining discrete curvature features provided in an embodiment of the present application;

[0033] Fig. 9 A schematic diagram of comparison of a local geodesic area before and after clipping provided in an embodiment of the present application;

[0034] Fig.10 A schematic diagram of the structure of a quality evaluation device for a three-dimensional texture grid provided in an embodiment of the present application;

[0035] Fig.11 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0036] The embodiments of the present application are described below in conjunction with the drawings in the present application. It should be understood that the implementation methods described below in conjunction with the drawings are exemplary descriptions for explaining the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions of the embodiments of the present application.

[0037] It will be understood by those skilled in the art that, unless specifically stated, the singular forms "one", "an", and "the" used herein may also include plural forms. It should be further understood that the terms "including" and "comprising" used in the embodiments of the present application refer to that the corresponding features can be implemented as the presented features, information, data, steps, operations, elements and / or components, but do not exclude the implementation as other features, information, data, steps, operations, elements, components and / or combinations thereof supported by the technical field. It should be understood that when an element is "connected" or "coupled" to another element, the element may be directly connected or coupled to the other element, or may refer to the element and the other element establishing a connection relationship through an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein indicates at least one of the items defined by the term, such as "A and / or B" indicates implementation as "A", or implementation as "A", or implementation as "A and B". "Multiple" refers to two or more.

[0038] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0039] See also Figure 1 , Figure 1 A schematic diagram of a system architecture provided in an embodiment of the present application. Figure 1 As shown, system 10 includes an encoding end 12, which can provide encoded video data to a decoding end 14, and the decoding end 14 can decode the encoded video data. The encoding end 12 and the decoding end 14 can include any of a variety of devices, including a desktop computer, a notebook computer, a tablet computer, a set-top box, an intelligent terminal (such as a so-called "smart" phone), a television, a camera, a display device, a digital media player, a video game console, or at least one of a video streaming device. In some cases, the encoding end 12 and the decoding end 14 are equipped for wireless communication. Therefore, the encoding end 12 and the decoding end 14 can be wireless communication devices. The technology described in the embodiments of the present disclosure can be applied to wireless and / or wired applications. The encoding end 12 is an exemplary video encoding device, that is, a device for encoding video data. The decoding end 14 is an exemplary video decoding device, that is, a device for decoding video data.

[0040] Figure 1The system 10 shown is only an example. The technology for encoding, decoding and processing video data can be performed by any digital video encoding and / or decoding device. In some examples, the technology can be performed by a video encoder / decoder, which is generally referred to as a "codec (CODEC)". The encoding end 12 and the decoding end 14 are examples of such codec devices, wherein the encoding end 12 generates encoded video data to send to the decoding end 14. In some examples, the encoding end 12 and the decoding end 14 operate in a substantially symmetrical manner, so that each of the encoding end 12 and the decoding end 14 includes a video encoding and decoding component. Therefore, the system 10 can support one-way or two-way video transmission between the encoding end 12 and the decoding end 14, such as for video streaming, video playback, video broadcasting or video telephony.

[0041] exist Figure 1 In the example of , the encoding end 12 includes a video source 121, a video encoder 122, and an output interface 123. The decoding end 14 includes an input interface 141, a video decoder 142, and a display device 143. In other examples, the encoding end 12 and the decoding end 14 include other components or arrangements. For example, the encoding end 12 can receive video data from an external video source (such as an external camera). Similarly, the decoding end 14 can be connected to an external display device instead of including an integrated display device.

[0042] Video source 121 is a video data source. Video data may include a series of pictures. Video source 121 may include a video capture device (such as a camera), a video archive containing previously captured video and / or a video feed interface for receiving video data from a video content provider. In some examples, video source 121 generates video data based on computer graphics, or a combination of live video, archived video and computer-generated video. In each case, the captured, pre-captured or computer-generated video may be encoded by video encoder 122.

[0043] In this example, optionally, when the encoding end 12 needs to encode the video, the reference grid can be encoded in different encoding methods to obtain a distorted grid, and then the quality of the distorted grid obtained by each encoding method is evaluated through the technical solution of the embodiment of the present application, and then one of the encoding methods can be selected for encoding based on the result of the quality evaluation. Then, the encoding end 12 sends a bit stream to the decoding end 14, and the bit stream includes the video to be decoded and the encoding method used by the encoding end 12, and the decoding end 14 can decode based on the decoding method corresponding to the encoding method.

[0044] In view of at least one of the above-mentioned technical problems or areas that need improvement in the point cloud-based evaluation method, the present application proposes a quality evaluation method and related device for a three-dimensional texture map grid, which obtains a distorted grid to be evaluated and a reference grid corresponding to the distorted grid; samples points in at least one grid to be processed to obtain a target point set, and at least one grid to be processed is at least one of the reference grid and the distorted grid; for each target point in the target point set, based on the distance between the points, determines the first nearest neighbor point of the target point from the reference grid, and determines the second nearest neighbor point of the target point from the distorted grid. Near point; for each target point, the first nearest neighbor point of the target point is taken as the region center point, and the first local geodesic region corresponding to the target point is determined from the reference grid, and the second nearest neighbor point of the target point is taken as the region center point, and the second local geodesic region corresponding to the target point is determined from the distorted grid; for each target point, the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point is determined; based on the similarity corresponding to each target point, the quality evaluation result of the distorted grid is determined, which may not require the topological structure of the grid, thereby realizing the applicability of the grid quality evaluation.

[0045] Among them, the method provided in the embodiment of the present application may involve artificial intelligence (AI) technology, for example, using artificial intelligence to process the texture map to obtain a reference grid, and for another example, using artificial intelligence to process the reference grid to obtain a distorted grid.

[0046] With the research and advancement of artificial intelligence technology, artificial intelligence technology has been studied and applied in many fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, driverless cars, autonomous driving, drones, digital twins, virtual humans, robots, artificial intelligence generated content (AIGC), conversational interaction, smart medical care, smart customer service, game AI, etc. It is believed that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0047] Artificial intelligence is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0048] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. Basic artificial intelligence technologies generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, pre-trained model technology, operation / interaction systems, mechatronics, etc. Among them, pre-trained models are also called large models and basic models. After fine-tuning, they can be widely used in downstream tasks in various major directions of artificial intelligence. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0049] Optionally, the technical solution of the embodiment of the present application involves cloud technology. For example, the solution of the embodiment of the present application can be executed by a server or a user terminal, wherein the server can be a cloud server, and the data processing involved in the implementation of the solution can be implemented based on cloud technology, and the data storage involved in the implementation process can adopt cloud storage. For example, the grid and the grid data can adopt cloud storage.

[0050] Among them, cloud technology is a general term for network technology, information technology, integration technology, management platform technology, application technology, etc. based on the cloud computing business model application, which can form a resource pool and be used on demand, flexibly and conveniently. Cloud computing technology will become an important support. Cloud storage is a new concept extended and developed from the concept of cloud computing. Distributed cloud storage system (hereinafter referred to as storage system) refers to a storage system that uses cluster applications, grid technology, and distributed storage file systems to bring together a large number of different types of storage devices (storage devices are also called storage nodes) in the network through application software or application interfaces to work together and provide data storage and business access functions to the outside world.

[0051] The following describes several exemplary embodiments to illustrate the technical solutions of the embodiments of the present application and the technical effects produced by the technical solutions of the present application. It should be noted that the following embodiments can refer to, draw on or combine with each other, and the same terms, similar features and similar implementation steps in different embodiments will not be described repeatedly.

[0052] See also Figure 2 , Figure 2A flow chart of a quality evaluation method for a three-dimensional texture grid provided in an embodiment of the present application. The quality evaluation method of this embodiment may be a full-reference objective quality evaluation method, and the objective quality evaluation algorithm refers to the use of a computer program instead of the human eye to evaluate the quality of the media. The method of this embodiment may be executed by an electronic device, and the electronic device is one or more of various desktop computers, laptops, smart phones, tablet computers, Internet of Things devices, portable wearable devices, or immersive image display devices. The Internet of Things device may be one or more of a smart speaker, a smart TV, a smart air conditioner, or a smart car-mounted device. The portable wearable device may be one or more of a smart watch, a smart bracelet, or a head-mounted device. Immersive image display devices include, but are not limited to, augmented reality (AR) devices, virtual reality (VR) devices, and the like. The electronic device can be used as an encoding end to execute the method of the embodiment of the present application, such as Figure 2 The quality evaluation method of the three-dimensional texture map shown includes:

[0053] S210: Obtain a distorted grid to be evaluated and a reference grid corresponding to the distorted grid.

[0054] Among them, the reference mesh and the distorted mesh are three-dimensional meshes corresponding to the texture map. In this embodiment, the distorted mesh can be obtained by compressing the reference mesh. The three-dimensional mesh (Mesh) data is composed of a series of polygonal patches connected by points with clear connection relationships in three-dimensional space, and contains some optional additional attribute information, such as point normal vectors and texture map coordinates. The most widely used in industrial applications and manufacturing industries is the colored three-dimensional mesh with texture map coordinates and texture maps.

[0055] In this embodiment, the three-dimensional grid includes but is not limited to point information, face information, normal vector information, texture map coordinate information and texture map information. The point information is an M×3 matrix V, where M represents the number of points, and each point contains three-dimensional coordinates in space, such as the floating-point coordinates of the k-th point are represented as (xk, yk, zk), k=0, ..., M-1. The texture map coordinate information is a Q×2 matrix UV, where Q represents the number of points, and each row represents the texture map coordinates on the texture map, such as the floating-point coordinates of the i-th point are represented as (u i , v i ), i = 0, ..., Q-1. The normal vector is a Y × 3 matrix Nr, each row represents a normal vector coordinate, such as the floating point coordinate of nr-th is expressed as (nrx nr ,nry nr ,nrz nr), nr = 0, ..., Nr-1. The face information is a matrix F of N × 3 / 6 / 9, where N represents the number of faces. Each face contains at least a point index. For example, the integer type representation of the j-th face is (A j , B j , C j ), A j , B j , C j ∈{k|1≤k≤M} and A j ≠B j ≠C j , the dimension of F is N×3; when each face contains not only the face index but also the texture attribute information of the point represented by each index, the integer type representation of the j-th face is (A j / uv1 j , B j / uv2 j , C j / uv3 j ), A j , B j , C j ∈{k|1≤k≤M},A j ≠B j ≠C j , uv1 j , uv2 j , uv3 j ∈{i|1≤i≤Q}, the dimension of F is N×6; when each face contains not only the face index and the texture attribute information of the point represented by each index, but also the normal vector information of each point, the integer type representation of the j-th face is (A j / uv1 j / nr1 j , B j / uv2 j / nr2 j , C j / uv3 j / nr3 j ), A j , B j , C j ∈{k|1≤k≤M} and A j ≠B j ≠C j , uv1 j , uv2 j , uv3 j ∈{i|1≤i≤Q},nr1 j , nr2 j , nr3 j∈{nr|1≤nr≤Y}, the dimension of F is N×9. For the convenience of expression, the face matrix is ​​split according to the information type, namely the point index matrix F_v, the point texture information matrix F_uv and the point normal vector matrix F_nr, F_v=(A,B,C), F_uv=(uv1,uv2,uv3), F_nr=(nr1,nr2,nr3).

[0056] S220. Sample points in at least one to-be-processed grid to obtain a target point set, wherein the at least one to-be-processed grid is at least one of a reference grid and a distorted grid.

[0057] In this embodiment, points in the reference grid may be sampled to obtain a target point set; points in the distorted grid may be sampled to obtain a target point set; points in the reference grid and the distorted grid may also be sampled to obtain a target point set.

[0058] The target point set includes multiple target points, and the number of target points in the target point set is less than or equal to the total number of grid points in the sampled grid to be processed. Optionally, the number of target points in the target point set is less than the total number of grid points in the sampled grid to be processed, which can improve the efficiency of quality evaluation.

[0059] Optionally, sampling may be performed using sampling methods such as random sampling, farthest point sampling (FPS), etc., which are not limited here.

[0060] S230: For each target point in the target point set, based on the distance between points, determine a first nearest point of the target point from the reference grid, and determine a second nearest point of the target point from the distorted grid.

[0061] In this embodiment, the first nearest point of the target point can be determined from the reference grid based on the distance between the target point and each point in the reference grid; and the second nearest point of the target point can be determined from the distorted grid based on the distance between the target point and each point in the distorted grid. Specifically, the distance between any two points can be determined by the coordinates of each of the two points, so that the point closest to the target point is selected as the nearest point.

[0062] S240. For each target point, the first nearest neighbor point of the target point is used as the region center point, and a first local geodesic region corresponding to the target point is determined from the reference grid; and the second nearest neighbor point of the target point is used as the region center point, and a second local geodesic region corresponding to the target point is determined from the distorted grid.

[0063] In this embodiment, the first nearest point can be used as the center point of the first local geodesic area. Similarly, the second nearest point can be used as the center point of the second local geodesic area. Specifically, the first local geodesic area corresponding to a target point can represent the grid area in the reference grid that is most relevant to the target point. Similarly, the second local geodesic area corresponding to a target point can represent the grid area in the distorted grid that is most relevant to the target point.

[0064] Exemplarily, assuming that the target point set includes target point 1 and target point 2, for target point 1, the first nearest point and the second nearest point of target point 1 are determined from the reference grid, and then the first local geodesic area and the second local geodesic area corresponding to target point 1 are determined. Similarly, for target point 2, the first nearest point and the second nearest point of target point 2 are determined from the reference grid, and then the first local geodesic area and the second local geodesic area corresponding to target point 2 are determined.

[0065] S250: For each target point, determine the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point.

[0066] The similarity indicates the similarity between the first local geodesic area and the second local geodesic area. Specifically, the greater the similarity, the greater the similarity between the grid area corresponding to the target point in the reference grid and the grid area corresponding to the target point in the distorted grid, that is, the higher the matching degree of the grid area between the reference grid and the distorted grid.

[0067] S260: Determine a quality evaluation result of the distorted grid based on the similarity corresponding to each target point.

[0068] Exemplarily, assuming that the target point set includes target point 1 and target point 2, the quality evaluation result of the distorted mesh is determined based on the similarity corresponding to target point 1 and the similarity corresponding to target point 2. The quality evaluation result may include a quality evaluation score, and a higher quality evaluation score indicates that the distorted mesh is more similar to the reference mesh, which may indicate that the mesh compression effect is better.

[0069] The technical solution of this embodiment is as follows: by obtaining a distorted grid to be evaluated and a reference grid corresponding to the distorted grid, the reference grid and the distorted grid are three-dimensional grids corresponding to the texture map; sampling points in at least one grid to be processed to obtain a target point set, the at least one grid to be processed is at least one of the reference grid and the distorted grid; for each target point in the target point set, based on the distance between the points, determining the first nearest neighbor point of the target point from the reference grid, and determining the second nearest neighbor point of the target point from the distorted grid; for each target point, taking the first nearest neighbor point of the target point as the center point of the region, determining the first local geodesic region corresponding to the target point from the reference grid, and taking the second nearest neighbor point of the target point as the center point of the region, determining the second local geodesic region corresponding to the target point from the distorted grid; for each target point, determining the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point; determining the quality evaluation result of the distorted grid based on the similarity corresponding to each target point, the topological structure of the grid can be arbitrary, thereby improving the applicability of three-dimensional grid evaluation.

[0070] It should be noted that in this embodiment, the number of sampled points can be set as needed. In some cases of the examples, the more target points in the target point set, the more accurate the quality evaluation result, but the quality evaluation efficiency is correspondingly reduced, so a solution is needed that is compatible with the accuracy and efficiency of the quality evaluation result.

[0071] In a possible implementation, sampling points in at least one to-be-processed grid to obtain a target point set includes:

[0072] Acquire sampling quantity reference information; determine the sampling quantity based on the sampling quantity reference information; and sample points in at least one to-be-processed grid based on the sampling quantity to obtain a target point set, wherein the target point set includes the sampling number of target points.

[0073] In this embodiment, the sampling quantity reference information includes at least one of the following:

[0074] Mesh size information of the reference mesh or the distorted mesh; image information of a texture map corresponding to the reference mesh or the distorted mesh; hardware capability information of a terminal executing the method;

[0075] The grid size information may indicate the size of the reference grid or the distorted grid. Optionally, the grid size information may include a first length of the grid along a first direction, a second length along a second direction, and a third length along a third direction, wherein the first direction, the second direction, and the third direction are perpendicular to each other. Generally speaking, the larger the size of the reference grid or the distorted grid indicated by the grid size information, the larger the number of samples.

[0076] See also Figure 3 , Figure 3 A schematic diagram of a reference grid or a distorted grid provided in an embodiment of the present application.

[0077] Specifically, the floating-point coordinates of the k-th point of the input reference or distortion grid are expressed as (x k ,y k , z k ), k = 0, ..., K-1, K is the number of points in the grid, the coordinate point (x min ,y min , z min ) and (x max ,y max , z max ) is expressed as follows:

[0078] x min =min(x 0 , x 1 , …, x K-1 )

[0079] y min =min(y 0 ,y 1 , …, y K-1 )

[0080] z min =min(z 0 , z 1 , …, z K-1 )

[0081] x max =max(x 0 , x 1 , …, x K-1 )

[0082] y max =max(y 0 ,y 1 , …, y K-1 )

[0083] z max =max(z 0 , z 1 , …, z K-1 )

[0084] Function min(s 0 ,s 1 ,…,s K-1 ), max(s 0 ,s 1 ,…,s K-1 ) means taking the minimum and maximum values ​​of the current input.

[0085] Thus, we can get Figure 3 The reference grid or the distorted grid shown can then determine the number of samples given the grid size information of the reference grid or the distorted grid.

[0086] The image information of the texture map may include at least one of the color richness and irregularity of the object in the texture map. The more colors, the higher the color richness, and the larger the number of samples. The higher the irregularity, the larger the number of samples. Exemplarily, the number of samples corresponding to the irregular object is greater than the number of samples corresponding to the regular object. The hardware capability information of the terminal may indicate the data processing capability of the terminal, and the hardware capability information may include at least one of memory, number of cores, or operating frequency. Specifically, the stronger the data processing capability, the larger the number of samples.

[0087] In this embodiment, by obtaining sampling quantity reference information; determining the sampling quantity based on the sampling quantity reference information; sampling from points in at least one grid to be processed based on the sampling quantity, a target point set is obtained, and the target point set includes the sampling number of target points, and the sampling quantity reference information includes at least one of the following: grid size information of the reference grid; image information of the texture map corresponding to the reference grid; hardware capability information of the terminal executing the method, the accuracy and efficiency of the quality evaluation results can be compatible.

[0088] The following embodiment describes how to obtain a local geodesic area based on any of the above embodiments. In this embodiment, for each target point, the first local geodesic area and the second local geodesic area corresponding to the target point are obtained by:

[0089] Determine each neighboring point in the reference grid that is connected to the first nearest neighbor, and use the area corresponding to the first nearest neighbor and each neighboring point of the first nearest neighbor as the first local geodesic area, wherein the first nearest neighbor is the area center point of the first local geodesic area;

[0090] Each neighboring point in the distorted grid that is connected to the second nearest neighboring point is determined, and the area corresponding to the second nearest neighboring point and each neighboring point of the second nearest neighboring point is used as the second local geodesic area, wherein the second nearest neighboring point is the area center point of the second local geodesic area.

[0091] It should be noted that the neighboring point that is connected to the first nearest neighbor may be a neighboring point directly connected to the first nearest neighbor or a neighboring point indirectly connected to the first nearest neighbor, and this is not limited here. Similarly, the neighboring point that is connected to the second nearest neighbor may be a one-hop neighbor directly connected to the second nearest neighbor or a multi-hop neighbor indirectly connected to the second nearest neighbor, and this is not limited here. Specifically, if a multi-hop neighbor is indirectly connected to one of the nearest neighbors, it means that there are other grid points before the multi-hop neighbor connected to the nearest neighbor, such as a one-hop neighbor.

[0092] See also Figure 4 . Figure 4 is a schematic diagram of a local geodesic area determined based on one-step neighboring points provided in an embodiment of the present application. Figure 4 The local geodesic region shown is based on the nearest neighbor and the one-step neighbor directly connected to the nearest neighbor.

[0093] like Figure 4 The local geodesic area shown may be a three-dimensional solid area.

[0094] In this embodiment, the first local geodesic area and the second local geodesic area can be connected by Figure 4 The method is obtained in the following way, which is not limited here.

[0095] The following embodiment describes how to determine the similarity between the first local geodesic area and the second local geodesic area corresponding to each target point based on any of the above embodiments.

[0096] For each target point, determining the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point includes:

[0097] For each area in the first local geodesic area and the second local geodesic area corresponding to the target point, determine the area feature of the area based on the area information of the area, where the area feature includes at least one of a smoothness feature, a texture map feature or a discrete curvature feature of the area;

[0098] Based on the similarity between the regional features of the first local geodesic area and the regional features of the second local geodesic area, the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point is determined.

[0099] The region information may include point information, surface information, normal vector information, texture map coordinate information, texture map information, etc. The smoothness feature is used to represent the smoothness of the region. The texture map feature is used to represent the texture of the region. The texture may include color, grayscale, etc. The discrete curvature feature is used to represent the curvature of the region.

[0100] In this embodiment, the similarity between regions can be determined based on the similarity between the regional features of the regions. Specifically, if the regional features include one of a smoothness feature, a texture map feature, or a discrete curvature feature, the similarity between regions can be determined based on the similarity between the smoothness features of the regions, or based on the similarity between the texture map features of the regions, or based on the similarity between the discrete curvature features. If the regional features include a smoothness feature, a texture map feature, and a discrete curvature feature, the similarity between the smoothness features, the similarity between the texture map features, and the similarity between the discrete curvature features can be calculated respectively, and then the similarity between the regions can be determined based on the similarity between the features.

[0101] The technical solution of this embodiment calculates the similarity between smoothness features, the similarity between texture map features, and the similarity between discrete curvature features respectively, and then determines the similarity between regions based on the similarity between each feature. Multiple features can be combined to determine the similarity, thereby improving the accuracy of the similarity between regions and thus improving the accuracy of quality evaluation.

[0102] The following embodiments respectively illustrate how to obtain a smoothness feature, how to obtain a texture map feature, and how to obtain a discrete curvature feature.

[0103] First, how to obtain the smoothness feature is described.

[0104] See also Figure 5 , Figure 5 A schematic diagram of a process for obtaining a smoothness feature provided in an embodiment of the present application. For each region, the smoothness feature of the region can be obtained by Figure 5 The method shown is obtained, as Figure 5 The methods shown include:

[0105] S510: For all points in the region, based on the coordinates of the points and the connection relationship between the points, determine the correlation between any two points among all the points.

[0106] In this embodiment, the correlation may include the Euclidean distance between two points.

[0107] According to the point and surface information in the local geodesic area, the edge weight parameters of the two endpoints of each edge of the surface are calculated. The calculation is preferably performed according to the Gaussian function:

[0108]

[0109] Where a and b are two points in the local area, V(a) and V(b) are their spatial coordinates, ||V(a)-V(b)|| 2 2 Represents the square of the second norm, that is, the Euclidean distance between two points. σ is a parameter that can be customized.

[0110] S520: Take each of all the points as a node in the graph structure corresponding to the region, take the correlation between any two points as the weight of the edge between the two corresponding nodes, and obtain an adjacency matrix corresponding to the graph structure.

[0111] Using the above method, for a local geodesic region with N points, we can obtain an adjacency matrix W with a dimension of N×N, whose elements W i，j It represents the spatial dependency between two points with indices i and j, that is, any element in the adjacency matrix represents the correlation between the two points.

[0112] S530: Based on the adjacency matrix, obtain a Laplacian matrix corresponding to the graph structure.

[0113] The Laplacian matrix of this embodiment can be a normalized Laplacian matrix or a normalized Laplacian matrix. Based on the adjacency matrix W, the degree matrix D and the Laplacian matrix L can be calculated: D is a diagonal matrix with a dimension of N×N, and the elements of each row are the sum of the corresponding row elements of the adjacency matrix, and L=DW. The normalized Laplacian matrix is * represents matrix multiplication.

[0114] S540 , obtaining point texture information of each point among all the points, and obtaining point features of each point among all the points based on the point texture information.

[0115] The point texture map information may include but is not limited to color information. The color information may be RGB color or YUV color, etc., which is not limited here. The point feature may be RGB value or YUV value, etc.

[0116] S550: Use the point feature of each point as the point signal of the node corresponding to the point, and obtain the graph signal corresponding to the graph structure based on the point signals of all points.

[0117] In this embodiment, taking the YUV color space as an example, a graphic signal matrix Fs with a dimension of N×3 can be obtained, and the three dimensions correspond to the Y, U, and V color information of N points respectively.

[0118] S560: Based on the Laplacian matrix corresponding to the graph structure and the graph signal, obtain the smoothness feature of the region.

[0119] The smoothness of the graph signal can be calculated as follows: S i =Fs i ′*Ln*Fs i , where Fs i represents the i-th dimension of the image signal matrix Fs, i=1, 2 or 3, ' represents the matrix transpose. i represents the three color channels of RGB color or the three color channels of YUV color.

[0120] Specifically, in the related art, point texture information is not used to evaluate the quality of the texture map grid.

[0121] In this embodiment, the point texture information of each point among all points is obtained, and the point feature of each point among all points is obtained based on the point texture information. Then, the point feature of each point is used as the point signal of the node corresponding to the point, and based on the point signals of all points, the graph signal corresponding to the graph structure is obtained, that is, the graph signal is obtained by referring to the point texture information.

[0122] In this embodiment, the smoothness of the image signal of the region can be directly used as the smoothness of the region.

[0123] However, in some exemplary cases, the regional parameters of different regions may be different, for example, the regional sizes of the regions may be different. In this case, if the smoothness of the image signal of the region is directly used as the smoothness of the region for quality evaluation, the accuracy of the quality evaluation may be affected to a certain extent. Therefore, the following embodiment, based on any of the above embodiments, describes how to further improve the accuracy of the quality evaluation of the texture map grid.

[0124] In a possible implementation, based on the Laplacian matrix and the graph signal corresponding to the graph structure, the smoothness feature of the region is obtained, including:

[0125] Based on the Laplacian matrix corresponding to the graph structure and the graph signal, the smoothness characteristics of the graph signal are obtained;

[0126] For each patch included in the region, determine pixel information of the enclosing region corresponding to the patch in the texture map, where the pixel information includes the number of pixel points in the enclosing region;

[0127] Determine the total number of pixels in the texture map of the region based on the number of pixels corresponding to each patch in the region;

[0128] Based on the smoothness feature and the total number of the image signal, the smoothness feature of the region is obtained, and the smoothness feature of the region is positively correlated with the ratio, wherein the ratio is the ratio of the smoothness feature of the image signal to the total number.

[0129] Among them, the patch can be a surface formed by multiple points connected in sequence in a local area. The enclosing area represents a three-dimensional area where the patch is mapped in the texture map. Optionally, the patch of this embodiment can be a polygonal patch, such as a triangular patch, which is not limited here. Specifically, the cross-sectional shape of the enclosing area is consistent with the shape of the patch.

[0130] See also Figure 6 , Figure 6 A schematic diagram of an enclosing area corresponding to a patch in a texture map provided in an embodiment of the present application.

[0131] like Figure 6 As shown, taking a triangular patch as an example, the enclosing area corresponding to the patch in the texture map may be a triangular prism, and the cross-sectional shape of the triangular prism is a triangle.

[0132] In this embodiment, a region may include multiple patches, and each patch has a corresponding enclosing region in the texture map. Based on the number of pixels corresponding to each patch in a region, the total number of pixels corresponding to the region can be determined.

[0133] Then, the local geodesic region average smoothness characteristic can be F 1i =S i / P n .

[0134] The technical solution of this embodiment is as follows: for each facet included in the area, determining the pixel information of the enclosing area corresponding to the facet in the texture map, the pixel information includes the number of pixel points in the enclosing area; based on the number of pixel points corresponding to each facet in the area, determining the total number of pixel points in the area in the texture map; based on the smoothness feature and the total number of the image signal, obtaining the smoothness feature of the area, the smoothness feature of the area is positively correlated with the inverse of the total number and the smoothness feature of the image signal, respectively. That is, taking into account the situation that the regional parameters between different areas may be different, the accuracy of the smoothness feature can be improved, thereby improving the accuracy of the mesh quality evaluation.

[0135] In a possible implementation, for each patch, determining pixel information of an enclosing area corresponding to the patch in the texture map includes:

[0136] Convert the texture map UV coordinates corresponding to each endpoint of the patch into texture map Cartesian coordinates;

[0137] Based on the Cartesian coordinates corresponding to each endpoint of the patch, an enclosing area in the texture map corresponding to the patch is obtained;

[0138] Extract pixel information from the enclosing area corresponding to the patch.

[0139] Among them, UV coordinates mean that all image files are a two-dimensional plane. In this embodiment, the texture map UV coordinates corresponding to each endpoint of the patch are converted into texture map Cartesian coordinates, and then the enclosed area can be obtained, so as to extract pixel information from the enclosed area. Specifically, the pixel information of this embodiment includes but is not limited to the number of pixels and the pixel value of each pixel. The pixel value can be a binary value, an RGB value, or a YUV value, which is not limited here.

[0140] Next, the following embodiments illustrate how to obtain texture features based on any of the above embodiments. Figure 7 , Figure 7 A schematic diagram of a process for obtaining texture map features provided in an embodiment of the present application. For each region, the texture map features of the region can be obtained by Figure 7 The method shown is obtained, as Figure 7 The methods shown include:

[0141] S710: For each patch included in the region, determine pixel information of an enclosing region corresponding to the patch in the texture map, where the pixel information includes a pixel value of each pixel point in the enclosing region.

[0142] The pixel value may be a binary value, an RGB value, or a YUV value, which is not limited here.

[0143] S720: For each patch included in the region, determine a pixel feature corresponding to the patch based on the pixel values ​​of each pixel point corresponding to the patch.

[0144] The pixel feature corresponding to a patch may be the pixel average value of all the pixels in the patch or the pixel value variance of all the pixels in the patch, which is not limited here.

[0145] S730: Fuse the pixel features corresponding to each patch in the region to obtain the texture map features of the region.

[0146] The fusion may be to calculate the average value of the pixel features corresponding to each slice in a region, or to take the maximum value or the minimum value, which is not limited here.

[0147] The technical solution of this embodiment is to determine the pixel information of the enclosing area corresponding to the patch in the texture map, the pixel information includes the pixel value of each pixel point in the enclosing area, and for each patch included in the area, determine the pixel feature corresponding to the patch based on the pixel value of each pixel point corresponding to the patch, and fuse the pixel features corresponding to each patch in the area to obtain the texture map feature of the area, which can further improve the accuracy of determining the texture map feature, and further improve the accuracy of mesh quality evaluation.

[0148] Optionally, the pixel feature may be one or more, which is not limited here.

[0149] In a possible implementation, the pixel value of a pixel includes the pixel values ​​of multiple color channels of the pixel, the pixel feature includes the pixel feature of each color channel in the multiple color channels, and the pixel features corresponding to each patch of the region are fused to obtain the texture map feature of the region, including:

[0150] For each color channel, the pixel features of each patch in the region corresponding to the color channel are fused to obtain the texture map features corresponding to the color channel.

[0151] Wherein, the texture map features of the region include texture map features corresponding to each channel in multiple color channels. The multiple color channels may be R channel, G channel and B channel, or may be Y channel, U channel and V channel. In an embodiment, for each color channel, there is a corresponding pixel feature. Wherein, the pixel feature corresponding to each color channel may include a pixel average value or a pixel value variance, which is not limited here. Exemplarily, the pixel average value Mij and the pixel value variance Vij under the j color channel. The j color channel represents one of the color channels.

[0152] In a possible implementation, the fusion is performed based on the pixel features corresponding to each patch in the region, and the average value of the pixel features corresponding to each patch can be directly calculated.

[0153] However, in some example situations, the situations between different patches may be different. If the average value is directly calculated, the texture map features may not be accurate enough, which may lead to inaccurate quality evaluation of the mesh. Therefore, the following embodiment describes how to further improve the quality evaluation of the mesh based on any of the above embodiments.

[0154] In a possible implementation, the pixel information also includes the number of pixels in the surrounding area, and the pixel features corresponding to each patch in the area are fused to obtain the texture map features of the area, including:

[0155] For each patch included in the region, determine a weight corresponding to the patch based on the number of pixels in the enclosing area corresponding to the patch, wherein the weight corresponding to the patch is positively correlated with the number of pixels corresponding to the patch;

[0156] Based on the weight corresponding to each patch and the corresponding pixel features of each patch, a weighted sum is performed to obtain the texture map features of the area.

[0157] For example, if there are a number of patches Fm, the number of pixels of the i-th patch is pi, the average value Mij and variance Vij of the texture information under the j color channel, then the weight Use this weight to weight the average and variance of the texture information of each patch to obtain the color channel texture map feature of the local geodesic region j

[0158] The technical solution of this embodiment can improve the accuracy of texture map features and thus improve the accuracy of mesh quality evaluation by determining the weight corresponding to the patch based on the number of pixels in the enclosing area corresponding to the patch and then performing weighted summation.

[0159] The following embodiments illustrate how to obtain discrete curvature features based on any of the above embodiments. Figure 8 , Figure 8 A schematic diagram of a process for obtaining discrete curvature features provided in an embodiment of the present application. For each region, the discrete curvature features of the region can be obtained by Figure 8 The method shown is obtained, as Figure 8 The methods shown include:

[0160] S810: For all points in the region, based on the coordinates of the points and the connection relationship between the points, determine the correlation between any two points among all the points.

[0161] S820. Take each of all the points as a node in the graph structure corresponding to the region, take the correlation between any two points as the weight of the edge between the two corresponding nodes, and obtain an adjacency matrix corresponding to the graph structure, wherein any element in the adjacency matrix represents the correlation between two points.

[0162] Among them, S810 and S820 may refer to the description of any of the above embodiments, and will not be described in detail here.

[0163] S830. Determine the cotangent Laplace-Beltrami operator of the region center point corresponding to the region based on the adjacency matrix corresponding to the graph structure.

[0164] Optionally, the cotangent Laplace-Beltrami operator of the region center point of a region can be calculated by the following formula:

[0165]

[0166] Where i represents the row index of the region center in W, A represents the local average area (such as the Voronoi area), N(i) represents the index set of all points that have a clear connection relationship with the region center in the neighborhood matrix W, and α ij and βij They represent the two diagonals of the side formed by points i and j, cot represents the cotangent operation, and V(i) and V(j) are their spatial coordinates.

[0167] S840: Obtain the area and normal vector of each facet of the region, and determine the normal vector of the region center point corresponding to the region based on the area and normal vector of each facet of the region.

[0168] Among them, an optional method is to calculate the normal vector Fnor of all patches in the local geodesic area i and area Fsize i , and calculate the area weight coefficient of each patch: Based on the above results, the point normal vector of the center point of the region can be obtained:

[0169]

[0170] S850: Determine a discrete curvature feature of the region based on a normal vector and a cotangent Laplace-Beltrami operator of a region center point of the region.

[0171] Optional, local geodesic region discrete curvature features

[0172] The following embodiment describes how to calculate the similarity between the first local geodesic region and the second local geodesic region and how to determine the quality evaluation result based on any of the above embodiments.

[0173] Taking the regional features including at least one of the smoothness feature, the texture average feature (also called the pixel average feature), the texture variance feature (also called the pixel value variance feature) or the discrete curvature feature as an example, for each target point, for any target point, a pair of local geodesic regions can be obtained from the reference grid and the distorted grid based on it, and a set of features can be obtained:

[0174]

[0175] The superscript r represents that the feature is extracted from the reference local geodesic region, and the superscript d represents that the feature is extracted from the distorted local geodesic region. F1, F2, and F3 may have calculation results corresponding to multiple color channels. For each pair of features (with the same subscript number), an optional method is to calculate the similarity value between the two features of each pair of features using the following method:

[0176]

[0177] Where T represents a small non-zero constant that prevents the denominator from being zero. It is a number between 0 and 1, and the larger the value, the higher the similarity.

[0178] Specifically, the two features of a feature pair come from the first local geodesic area and the second local geodesic area respectively, and other conditions are the same, such as the same color channel, such as both are pixel averages, etc., which are not limited here.

[0179] Then, all feature similarities are pooled to obtain the similarity of this pair of local geodesic regions, and an adaptive weighting method can be used. An optional method is:

[0180]

[0181] Among them, mean[*] represents the mean value.

[0182] Optionally, for the target point set KP∈R ks×3 Each target point in has a corresponding similarity. Based on the similarity corresponding to each target point, the quality evaluation result of the distorted grid is determined, which may include:

[0183] Pooling the similarities corresponding to all target points can obtain the objective quality score of the distorted grid. An optional method is to calculate the mean:

[0184]

[0185] It can be understood that if the smoothness feature, texture average feature (also called pixel average feature), texture variance feature (also called pixel value variance feature) and discrete curvature feature are considered simultaneously for quality evaluation, it is equivalent to considering the geometric features and texture features of the grid at the same time, which can improve the accuracy of quality evaluation.

[0186] In some cases of the examples, there are some duplicate points, unindexed points, duplicate faces or invalid faces in the reference grid and the distorted grid, and these interferences may cause the quality evaluation to fail or the quality evaluation of the grid to be inaccurate. Therefore, the following embodiments, based on any of the above embodiments, illustrate how to improve the success rate and accuracy of the grid quality evaluation.

[0187] In a possible implementation manner, before sampling the points in at least one to-be-processed grid, the method further includes:

[0188] Perform grid purification on each to-be-processed grid to obtain a purified to-be-processed grid, wherein the network purification includes at least one of the following:

[0189] Remove duplicate points from the processed grid;

[0190] Delete the unindexed points in the mesh to be processed, where the unindexed points refer to points on any patch that does not belong to the mesh to be processed;

[0191] Remove duplicate faces from the processed mesh;

[0192] Deleting invalid faces in the mesh to be processed, where invalid faces refer to faces that include the same points or faces that include a number of points that is not greater than a preset number, where the preset number is an integer not less than 2;

[0193] Sampling points in at least one grid to be processed includes:

[0194] At least one point in the cleaned processing grid is sampled.

[0195] In this embodiment, deduplication of points in the to-be-processed mesh may be performed by searching for multiple duplicate points in the to-be-processed mesh and retaining one of the multiple duplicate points. Deleting unindexed points in the to-be-processed mesh may be performed by searching for unindexed points in the to-be-processed mesh and deleting the unindexed points found, wherein the unindexed points do not belong to any facet of the to-be-processed mesh. Optionally, if a point v∈R 3 , v∈V but , v is an unindexed point; if a point v∈R 3 , v∈V but there are multiple rows in V that are the same as v, and v is a duplicate point.

[0196] In this embodiment, deduplication of faces in the to-be-processed mesh may be performed by searching for multiple duplicate faces in the to-be-processed mesh and retaining one of the multiple duplicate faces. Deleting invalid faces in the to-be-processed mesh may be performed by searching for invalid faces in the to-be-processed mesh and deleting the invalid faces, wherein an invalid face indicates that it includes the same points or the number of points included is not greater than a preset number, and the preset number is an integer not less than 2. Optionally, the face index information needs to satisfy A j , B j , C j ∈{k|1≤k≤M} and A j ≠B j ≠C j The mesh to be processed may have faces that do not meet the above requirements, which are recorded as invalid faces, and there may be multiple faces with the same face information, which are recorded as equivalent faces.

[0197] It should be noted that after mesh purification, the mesh information of the mesh to be processed needs to be updated.

[0198] Specifically, if mesh purification is required for points and faces, mesh purification can be performed on faces first, and mesh information can be updated. Then, the faces to be purified can be found in the to-be-processed mesh after the mesh information is updated.

[0199] It can be understood that the above mesh purification process is performed cyclically, and when there are no duplicate points and unindexed points, the mesh purification can be considered completed.

[0200] In some cases of the examples, the mesh quality evaluation efficiency is low due to the large size of the first local geodesic area and the second local geodesic area, or the size difference between the first local geodesic area and the second local geodesic area is too large, which leads to low mesh quality evaluation accuracy. Therefore, the following embodiments, based on any of the above embodiments, illustrate how to improve the mesh quality evaluation efficiency and mesh quality evaluation accuracy.

[0201] In a possible implementation, for each target point, before determining the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point, the method includes:

[0202] For the first local geodesic area, determine a first distance between a region center point of the area and each neighboring point of the region center point, fuse the first distances corresponding to the neighboring points, and obtain a first size reference value of the first local geodesic area;

[0203] For the second local geodesic area, determine a second distance between the area center point of the area and each neighboring point of the area center point, and fuse the second distances corresponding to the neighboring points to obtain a second size reference value of the second local geodesic area;

[0204] Based on the first size reference value and the second size reference value, at least one local area is clipped to obtain a first local geodesic area and a second local geodesic area after clipping, wherein at least one area includes a local geodesic area corresponding to a larger value of the first size reference value and the second size reference value;

[0205] For each target point, determining the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point includes:

[0206] For each target point, the similarity between the clipped first local geodesic region and the second local geodesic region corresponding to the target point is determined.

[0207] In this embodiment, the first distances corresponding to the adjacent points are fused by summing the first distances or calculating the average value, which is not limited here. Similarly, the second distances corresponding to the adjacent points are fused by summing the second distances or calculating the average value, which is not limited here. Specifically, at least one local area in the first local geodesic area or the second local geodesic area can be cropped according to actual conditions, which is not limited here.

[0208] It should be noted that the first distance and the second distance may be geodesic distances. Geodesic distance is the distance of the shortest path (executable path) between two points on a surface. In graph theory, Geodesic Distance is the distance of the shortest path between two nodes in a graph, which is different from Euclidean Distance, which is the shortest distance between two points, commonly used in geometric space.

[0209] The technical solution of this embodiment is to determine, for each local area in the first local geodesic area and the second local geodesic area, the distance between the area center point of the area and each neighboring point of the area center point, and then determine the size reference value of each local area, and then based on the first size reference value and the second size reference value, crop at least one local area to obtain the cropped first local geodesic area and the second local geodesic area, and then determine the similarity between the cropped first local geodesic area and the second local geodesic area corresponding to the target point, that is, refer to the size of the first local area and the second local area, which can improve the quality evaluation efficiency and accuracy of the grid.

[0210] In a possible implementation, based on the first size reference value and the second size reference value, at least one local area is clipped to obtain the clipped first local geodesic area and the second local geodesic area, including at least one of the following:

[0211] Determine a ratio between the first size reference value and the second size reference value, and based on the ratio, clip at least the local geodesic area corresponding to the larger value of the first size reference value and the second size reference value, so that the sizes of the clipped first local geodesic area and the second local geodesic area match each other;

[0212] The local geodesic regions whose corresponding distances in the first local geodesic region and the second local geodesic region are greater than the distance threshold are cropped so that the distances corresponding to the cropped local geodesic regions are less than or equal to the preset distance threshold.

[0213] In this embodiment, the local geodesic area corresponding to the larger value of the first size reference value and the second size reference value can be clipped so that the sizes of the first local geodesic area and the second local geodesic area after clipping match, thereby improving the accuracy of the quality evaluation of the grid. The size between the first local geodesic area and the second local geodesic area can be that the radius difference between the first sphere radius of the circumscribed sphere of the first local geodesic area and the second sphere radius of the circumscribed sphere of the second local geodesic area is less than the difference threshold.

[0214] Optionally, based on the reference value ratio between the first size reference value and the second size reference value, the local geodesic area corresponding to the larger value of the first size reference value and the second size reference value can be clipped. Specifically, the spherical radius of the circumscribed sphere of the clipped local area can be determined based on the reference value ratio, and then clipping can be performed according to the spherical radius.

[0215] See also Fig. 9 , Fig. 9 A schematic diagram of the comparison of a local geodesic area before and after clipping provided in an embodiment of the present application.

[0216] like Fig. 9 As shown, the spherical radius of the circumscribed sphere of the local area can be determined first, and then the local geodesic area can be clipped according to the spherical radius so that the spherical radius of the circumscribed sphere of the clipped local geodesic area is consistent with the determined spherical radius.

[0217] In this embodiment, the local geodesic areas whose corresponding distances in the first local geodesic area and the second local geodesic area are greater than a distance threshold can also be cropped so that the distances corresponding to the cropped local geodesic areas are less than or equal to the distance threshold, thereby improving the efficiency of the quality evaluation of the grid.

[0218] It should be noted that the distance threshold may be pre-set or calculated, and is not limited here.

[0219] In a possible implementation manner, threshold reference information is obtained, where the threshold reference information includes at least one of the following:

[0220] Mesh size information of the reference mesh or the distorted mesh; image information of a texture map corresponding to the reference mesh or the distorted mesh; hardware capability information of a terminal executing the method;

[0221] A distance threshold is determined based on the threshold reference information.

[0222] Among them, the grid size information can indicate the size of the reference grid or the distorted grid. Optionally, the grid size information can include a first length of the grid along the first direction, a second length along the second direction, and a third length along the third direction, respectively, and the first direction, the second direction, and the third direction are perpendicular to each other. Generally speaking, the larger the size of the reference grid or the distorted grid represented by the grid size information, the larger the distance threshold. Optionally, the distance threshold can be S times the maximum length of the first length, the second length, and the third length, where S is a positive number less than 1, for example, S times can be 0.0005 times.

[0223] The image information of the texture map may include at least one of the color richness and irregularity of the objects in the texture map. The more colors there are, the higher the color richness is, and the larger the number of samples is. The higher the irregularity is, the larger the distance threshold is. Exemplarily, the distance threshold corresponding to the irregular object is greater than the distance threshold corresponding to the regular object. The hardware capability information of the terminal may indicate the data processing capability of the terminal, and the hardware capability information may include at least one of memory, number of cores, or operating frequency. Specifically, the stronger the data processing capability is, the larger the distance threshold is.

[0224] In this embodiment, threshold reference information is obtained, and a distance threshold is determined based on the threshold reference information, wherein the threshold reference information includes at least one of the following: grid size information of a reference grid or a distorted grid; image information of a texture map corresponding to the reference grid or the distorted grid; and hardware capability information of a terminal executing the method; thereby being able to maintain both accuracy and efficiency of quality evaluation results.

[0225] In order to intuitively illustrate the quality assessment effect of the embodiments of the present application, the following embodiments provide some experimental data for illustration.

[0226] For comparison, we selected the following models for comparison: using the reference software provided by MPEG to calculate three image-based evaluation models: geo_PSNR, RGB_PSNR and YUV_PSNR, three point-based evaluation models: point-to-point (D1), point-to-plan (D2) and PCQM_PSNR, and using the MSU video quality evaluation tool to calculate three video-based evaluation models: PSNR, SSIM and VMAF. Three indicators are used to measure the performance of the model, namely Pearson product-moment correlation coefficient (PLCC), the Spearman rank-order correlation coefficient (SROCC) and root mean square error (RMSE). The larger the PLCC and SROCC, the better, and the smaller the RMSE, the better. The performance is as follows:

[0227]

[0228] It can be seen that this application shows the best performance in all three indicators.

[0229] See also Fig.10 , Fig.10 Schematic diagram of a three-dimensional texture grid quality evaluation device provided in an embodiment of the present application. The device 100 of this embodiment can be applied to electronic devices, such as Fig.10The device 100 shown includes a grid acquisition module 110, a sampling module 120, a neighbor point determination module 130, a local area determination module 140, a similarity determination module 150 and a quality evaluation module 160, wherein:

[0230] The grid acquisition module 110 is used to acquire the distorted grid to be evaluated and the reference grid corresponding to the distorted grid, where the reference grid and the distorted grid are three-dimensional grids corresponding to the texture map; the sampling module 120 is used to sample points in at least one grid to be processed to obtain a target point set, where the at least one grid to be processed is at least one of the reference grid and the distorted grid; the neighboring point determination module 130 is used to determine, for each target point in the target point set, the first nearest neighboring point of the target point from the reference grid based on the distance between the points, and the second nearest neighboring point of the target point from the distorted grid; the local area determination module Block 140 is used to, for each target point, use the first nearest neighbor point of the target point as the region center point to determine the first local geodesic region corresponding to the target point from the reference grid, and use the second nearest neighbor point of the target point as the region center point to determine the second local geodesic region corresponding to the target point from the distorted grid; similarity determination module 150 is used to determine, for each target point, the similarity between the first local geodesic region and the second local geodesic region corresponding to the target point; quality evaluation module 160 is used to determine the quality evaluation result of the distorted grid based on the similarities corresponding to each target point.

[0231] In a possible implementation, for each target point, the local area determination module 140 is further configured to:

[0232] Determine each neighboring point in the reference grid that is connected to the first nearest point, and use the area corresponding to the first nearest point and each neighboring point of the first nearest point as the first local geodesic area, wherein the first nearest point is the area center point of the first local geodesic area; determine each neighboring point in the distorted grid that is connected to the second nearest point, and use the area corresponding to the second nearest point and each neighboring point of the second nearest point as the second local geodesic area, wherein the second nearest point is the area center point of the second local geodesic area.

[0233] In a possible implementation, for each target point, when determining the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point, the similarity determination module 150 is configured to:

[0234] For each area in the first local geodesic area and the second local geodesic area corresponding to the target point, determine the regional features of the area based on the regional information of the area, the regional features including at least one of the smoothness features, texture map features or discrete curvature features of the area; based on the similarity between the regional features of the first local geodesic area and the regional features of the second local geodesic area, determine the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point.

[0235] In a possible implementation, for each region, the similarity determination module 150 is further configured to:

[0236] For all points in the area, based on the coordinates of the points and the connection relationship between the points, determine the correlation between all the points; take each point among all the points as a node in the graph structure corresponding to the area, take the correlation between the two points as the weight of the edge between the corresponding two nodes, and obtain the adjacency matrix corresponding to the graph structure, wherein any element in the adjacency matrix represents the correlation between the two points; based on the adjacency matrix, obtain the Laplacian matrix corresponding to the graph structure; obtain the point texture information of each point among all the points, and obtain the point feature of each point among all the points based on the point texture information; take the point feature of each point as the point signal of the node corresponding to the point, and obtain the graph signal corresponding to the graph structure based on the point signals of all the points; based on the Laplacian matrix and the graph signal corresponding to the graph structure, obtain the smoothness feature of the area.

[0237] In a possible implementation, when obtaining the smoothness feature of the region based on the Laplacian matrix and the graph signal corresponding to the graph structure, the similarity determination module 150 is used to:

[0238] Based on the Laplacian matrix and the graph signal corresponding to the graph structure, the smoothness feature of the graph signal is obtained; for each patch included in the area, the pixel information of the enclosing area corresponding to the patch in the texture map is determined, and the pixel information includes the number of pixel points in the enclosing area; based on the number of pixel points corresponding to each patch in the area, the total number of pixel points in the texture map of the area is determined; based on the smoothness feature and the total number of the graph signal, the smoothness feature of the area is obtained, and the smoothness feature of the area is positively correlated with the inverse of the total number and the smoothness feature of the graph signal, respectively.

[0239] In a possible implementation, for each region, the similarity determination module 150 is further configured to:

[0240] For each patch included in the area, determine the pixel information of the enclosing area corresponding to the patch in the texture map, the pixel information including the pixel value of each pixel point in the enclosing area; for each patch included in the area, determine the pixel feature corresponding to the patch based on the pixel value of each pixel point corresponding to the patch; the pixel features corresponding to each patch in the area are fused to obtain the texture map feature of the area.

[0241] In a possible implementation, the pixel feature includes the pixel feature of each color channel in the multiple color channels. When the similarity determination module 150 fuses the pixel features corresponding to the patches in the region to obtain the texture map feature of the region, it is further used to:

[0242] For each color channel, the pixel features of each patch in the region corresponding to the color channel are fused to obtain the texture map features corresponding to the color channel.

[0243] In a possible implementation, the pixel information also includes the number of pixels in the surrounding area. When the similarity determination module 150 fuses the pixel features corresponding to the patches in the area to obtain the texture map features of the area, it is used to:

[0244] For each patch included in the area, the weight corresponding to the patch is determined based on the number of pixel points in the enclosing area corresponding to the patch, wherein the weight corresponding to the patch is positively correlated with the number of pixel points corresponding to the patch; weighted summation is performed based on the weights corresponding to each patch and the corresponding pixel features corresponding to each patch to obtain the texture map features of the area.

[0245] In a possible implementation, for each patch, when determining the pixel information of the enclosing area corresponding to the patch in the texture map, the similarity determination module 150 is used to:

[0246] The texture map UV coordinates corresponding to each endpoint of the patch are converted into texture map Cartesian coordinates to obtain an enclosing area in the texture map corresponding to the patch; and pixel information is extracted from the enclosing area corresponding to the patch.

[0247] In a possible implementation, for each region, the similarity determination module 150 is used to:

[0248] For all points in the area, based on the coordinates of the points and the connection relationship between the points, determine the correlation between any two points among all the points; take each point among all the points as a node in the graph structure corresponding to the area, take the correlation between any two points as the weight of the edge between the corresponding two nodes, and obtain the adjacency matrix corresponding to the graph structure, wherein any element in the adjacency matrix represents the correlation between the two points; based on the adjacency matrix corresponding to the graph structure, determine the cotangent Laplace-Beltrami operator of the region center point corresponding to the area; obtain the area and normal vector of each facet of the area, and determine the normal vector of the region center point corresponding to the area based on the area and normal vector of each facet of the area; based on the normal vector of the region center point and the cotangent Laplace-Beltrami operator, determine the discrete curvature characteristics of the area.

[0249] In a possible implementation, the device 100 further includes:

[0250] The grid purification module is used to perform grid purification on each grid to be processed to obtain a purified grid to be processed, wherein the network purification includes at least one of the following:

[0251] Remove duplicate points from the processed grid;

[0252] Delete the unindexed points in the mesh to be processed, where the unindexed points refer to points on any patch that does not belong to the mesh to be processed;

[0253] Remove duplicate faces from the processed mesh;

[0254] Deleting invalid faces in the mesh to be processed, where invalid faces refer to faces that include the same points or faces that include a number of points that is not greater than a preset number, where the preset number is an integer not less than 2;

[0255] The sampling module 120 is used to sample points in at least one cleaned grid to be processed.

[0256] In a possible implementation, the device 100 further includes:

[0257] The cropping module is used to determine, for a first local geodesic area, a first distance between a region center point of the area and each neighboring point of the region center point, and fuse the first distances corresponding to the neighboring points to obtain a first size reference value of the first local geodesic area; for a second local geodesic area, determine a second distance between a region center point of the area and each neighboring point of the region center point, and fuse the second distances corresponding to the neighboring points to obtain a second size reference value of the second local geodesic area; based on the first size reference value and the second size reference value, crop at least one local area to obtain the cropped first local geodesic area and the second local geodesic area, wherein at least one area includes a local geodesic area corresponding to the larger value of the first size reference value and the second size reference value; the similarity determination module 150 is used to determine, for each target point, the similarity between the cropped first local geodesic area and the second local geodesic area corresponding to the target point.

[0258] In a possible implementation, when the cropping module performs cropping processing on at least one local area based on the first size reference value and the second size reference value to obtain the cropped first local geodesic area and the second local geodesic area, it is used for at least one of the following:

[0259] Determine a ratio between the first size reference value and the second size reference value, and based on the ratio, clip at least the local geodesic area corresponding to the larger value of the first size reference value and the second size reference value, so that the sizes of the clipped first local geodesic area and the second local geodesic area match each other;

[0260] The local geodesic regions whose corresponding distances in the first local geodesic region and the second local geodesic region are greater than the distance threshold are cropped so that the distances corresponding to the cropped local geodesic regions are less than or equal to the preset distance threshold.

[0261] The quality evaluation device for the three-dimensional texture map mesh of this embodiment can execute the quality evaluation method for the three-dimensional texture map mesh shown in the above-mentioned embodiments of the present application. The implementation principle is similar. The actions performed by each module in the device of each embodiment of the present application correspond to the steps in the method of each embodiment of the present application. For the detailed functional description of each module of the device, please refer to the description in the corresponding method shown in the previous text, which will not be repeated here.

[0262] An embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory, and the processor executes the above-mentioned computer program to implement the steps of the quality evaluation method of a three-dimensional texture grid.

[0263] In an alternative embodiment, an electronic device is provided, such as Fig.11 As shown, Fig.11 The electronic device 1100 shown includes: a processor 1101 and a memory 1103. The processor 1101 and the memory 1103 are connected, such as through a bus 1102. Optionally, the electronic device 1100 may also include a transceiver 1104, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception. It should be noted that in actual applications, the transceiver 1104 is not limited to one, and the structure of the electronic device 1100 does not constitute a limitation on the embodiments of the present application.

[0264] Processor 1101 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 1101 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0265] The bus 1102 may include a path to transmit information between the above components. The bus 1102 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 1102 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0266] The memory 1103 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), magnetic disk storage media, other magnetic storage devices, or any other medium that can be used to carry or store computer programs and can be read by a computer, without limitation herein.

[0267] The memory 1103 is used to store the computer program for executing the embodiment of the present application, and the execution is controlled by the processor 1101. The processor 1101 is used to execute the computer program stored in the memory 1103 to implement the steps shown in the above method embodiment.

[0268] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps and corresponding contents of the aforementioned method embodiment can be implemented.

[0269] The embodiment of the present application also provides a computer program product, including a computer program, which can implement the steps and corresponding contents of the aforementioned method embodiment when executed by a processor.

[0270] It should be understood that, although each operation step is indicated by arrows in the flowchart of the embodiment of the present application, the implementation order of these steps is not limited to the order indicated by the arrows. Unless clearly stated herein, in some implementation scenarios of the embodiment of the present application, the implementation steps in each flowchart can be performed in other orders according to demand. In addition, some or all of the steps in each flowchart may include multiple sub-steps or multiple stages based on actual implementation scenarios. Some or all of these sub-steps or stages may be executed at the same time, and each sub-step or stage in these sub-steps or stages may also be executed at different times respectively. In different scenarios of execution time, the execution order of these sub-steps or stages may be flexibly configured according to demand, and the embodiment of the present application does not limit this.

[0271] The above are only optional implementation methods for some implementation scenarios of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the technical concept of the scheme of the present application, other similar implementation methods based on the technical ideas of the present application are also within the protection scope of the embodiments of the present application.

Claims

1. A quality evaluation method for a three-dimensional texture grid, characterized in that: include: Acquire a distorted mesh to be evaluated and a reference mesh corresponding to the distorted mesh, wherein the reference mesh and the distorted mesh are three-dimensional meshes corresponding to a texture map; Sampling points in at least one to-be-processed grid to obtain a target point set, wherein the at least one to-be-processed grid is at least one of the reference grid and the distorted grid; For each target point in the target point set, based on the distance between points, determine a first nearest point of the target point from the reference grid, and determine a second nearest point of the target point from the distorted grid; For each of the target points, a first nearest neighbor point of the target point is used as a region center point, and a first local geodesic region corresponding to the target point is determined from the reference grid; and a second nearest neighbor point of the target point is used as a region center point, and a second local geodesic region corresponding to the target point is determined from the distorted grid; For each of the target points, determining a similarity between a first local geodesic region and a second local geodesic region corresponding to the target point; Based on the similarities corresponding to the target points, a quality evaluation result of the distorted grid is determined.

2. The method according to claim 1, characterized in that For each of the target points, the first local geodesic area and the second local geodesic area corresponding to the target point are obtained in the following manner: Determine each neighboring point in the reference grid that is connected to the first nearest point, and use the area corresponding to the first nearest point and each neighboring point of the first nearest point as the first local geodesic area; Each neighboring point in the distorted grid that is connected to the second nearest neighboring point is determined, and an area corresponding to the second nearest neighboring point and each neighboring point of the second nearest neighboring point is used as the second local geodesic area.

3. The method according to claim 1, characterized in that For each of the target points, determining the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point includes: For each area in the first local geodesic area and the second local geodesic area corresponding to the target point, determine the area feature of the area based on the area information of the area, the area feature including at least one of a smoothness feature, a texture map feature or a discrete curvature feature of the area; Based on the similarity between the regional features of the first local geodesic region and the regional features of the second local geodesic region, the similarity between the first local geodesic region corresponding to the target point and the second local geodesic region is determined. The similarity between the first local geodesic region corresponding to the target point and the second local geodesic region is determined.

4. The method according to claim 3, characterized in that: For each region, the smoothness characteristics of the region are obtained in the following way: For all points in the area, based on the coordinates of the points and the connection relationship between the points, determine the correlation between any two points in all the points; Taking each of all the points as a node in the graph structure corresponding to the region, taking the correlation between any two points as the weight of the edge between the corresponding two nodes, and obtaining an adjacency matrix corresponding to the graph structure, wherein any element in the adjacency matrix represents the correlation between two points; Based on the adjacency matrix, obtaining a Laplacian matrix corresponding to the graph structure; Acquire point texture information of each of the points, and obtain point features of each of the points based on the point texture information; The point feature of each point is used as the point signal of the node corresponding to the point, and based on the point signals of all points, a graph signal corresponding to the graph structure is obtained; Based on the Laplacian matrix and the graph signal corresponding to the graph structure, a smoothness feature of the region is obtained.

5. The method according to claim 4, characterized in that The obtaining of the smoothness feature of the region based on the Laplacian matrix and the graph signal corresponding to the graph structure includes: Based on the Laplacian matrix corresponding to the graph structure and the graph signal, obtaining a smoothness feature of the graph signal; For each patch included in the region, determine pixel information of the enclosing region corresponding to the patch in the texture map, wherein the pixel information includes the number of pixel points in the enclosing region; Determine the total number of pixels in the texture map for the region based on the number of pixels corresponding to each patch in the region; Based on the smoothness feature of the image signal and the total number, the smoothness feature of the region is obtained, and the smoothness feature of the region is positively correlated with the ratio, wherein the ratio is the ratio of the smoothness feature of the image signal to the total number.

6. The method according to claim 3, characterized in that For each region, the texture map features of the region are obtained in the following way: For each patch included in the region, determine pixel information of the enclosing region corresponding to the patch in the texture map, wherein the pixel information includes pixel values ​​of each pixel point in the enclosing region; For each patch included in the region, determining a pixel feature corresponding to the patch based on pixel values ​​of each pixel point corresponding to the patch; The pixel features corresponding to each patch in the area are fused to obtain the texture map features of the area.

7. The method according to claim 6, characterized in that The pixel value of a pixel point includes the pixel values ​​of multiple color channels of the pixel point, the pixel feature includes the pixel feature of each color channel in the multiple color channels, and the pixel features corresponding to each patch of the area are fused to obtain the texture map feature of the area, including: For each color channel, the pixel features of each patch in the region corresponding to the color channel are fused to obtain the texture map features corresponding to the color channel; The texture map features of the region include texture map features corresponding to each channel in the multiple color channels.

8. The method according to claim 6, characterized in that The pixel information also includes the number of pixels in the enclosed area. The pixel features corresponding to each patch in the area are fused to obtain the texture map features of the area, including: For each patch included in the region, determine a weight corresponding to the patch based on the number of pixels in the enclosing area corresponding to the patch, wherein the weight corresponding to the patch is positively correlated with the number of pixels corresponding to the patch; Based on the weight corresponding to each patch and the corresponding pixel features of each patch, a weighted sum is performed to obtain the texture map features of the area.

9. The method according to any one of claims 5 to 8, characterized in that: For each patch, determining pixel information of an enclosing area corresponding to the patch in the texture map includes: Convert the texture map UV coordinates corresponding to each endpoint of the patch into texture map Cartesian coordinates; Based on the Cartesian coordinates corresponding to each endpoint of the patch, obtaining an enclosing area in the texture map corresponding to the patch; Extract pixel information from the enclosing area corresponding to the patch.

10. The method according to claim 3, characterized in that For each region, the discrete curvature characteristics of the region are obtained in the following way: For all points in the area, based on the coordinates of the points and the connection relationship between the points, determine the correlation between any two points in all the points; Taking each of the points as a node in the graph structure corresponding to the region, taking the correlation between the two points as the weight of the edge between the two nodes, and obtaining an adjacency matrix corresponding to the graph structure, wherein any element in the adjacency matrix represents the correlation between the two points; Determine the cotangent Laplace-Beltrami operator of the region center point corresponding to the region based on the adjacency matrix corresponding to the graph structure; Obtain the area and normal vector of each facet of the region, and determine the normal vector of the region center point corresponding to the region based on the area and normal vector of each facet of the region; Based on the normal vector and the cotangent Laplace-Beltrami operator of the region center point of the region, the discrete curvature characteristics of the region are determined.

11. The method according to claim 1, characterized in that: Before sampling at least one point in the to-be-processed grid, the method further includes: Perform grid purification on each to-be-processed grid to obtain a purified to-be-processed grid, wherein the network purification includes at least one of the following: Deduplication of points in the grid to be processed; Deleting unindexed points in the to-be-processed mesh, wherein the unindexed points refer to points on any facet that does not belong to the to-be-processed mesh; Deduplication of faces in the mesh to be processed; Deleting invalid faces in the to-be-processed mesh, wherein the invalid faces refer to faces that include the same points or faces that include a number of points that is not greater than a preset number, where the preset number is an integer not less than 2; The sampling of points in at least one to-be-processed grid comprises: At least one cleaned point in the processing grid is sampled.

12. The method according to claim 1, characterized in that For each of the target points, before determining the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point, the method further comprises: For the first local geodesic area, determining a first distance between a region center point of the area and each neighboring point of the region center point, fusing the first distances corresponding to the neighboring points, and obtaining a first size reference value of the first local geodesic area; For the second local geodesic area, determine a second distance between a region center point of the area and each neighboring point of the region center point, and fuse the second distances corresponding to the neighboring points to obtain a second size reference value of the second local geodesic area; Based on the first size reference value and the second size reference value, at least one local area is clipped to obtain a first local geodesic area and a second local geodesic area after clipping, wherein the at least one area includes a local geodesic area corresponding to a larger value of the first size reference value and the second size reference value; For each of the target points, determining the similarity between the first local geodesic area and the second local geodesic area corresponding to the target point includes: For each of the target points, the similarity between the clipped first local geodesic region and the second local geodesic region corresponding to the target point is determined.

13. The method according to claim 12, characterized in that The clipping of at least one local area based on the first size reference value and the second size reference value to obtain the clipped first local geodesic area and the second local geodesic area includes at least one of the following: determining a ratio between the first size reference value and the second size reference value, and based on the ratio, clipping at least a local geodesic region corresponding to a larger value of the first size reference value and the second size reference value, so that sizes of the clipped first local geodesic region and the second local geodesic region match; The local geodesic regions whose corresponding distances in the first local geodesic region and the second local geodesic region are greater than the distance threshold are cropped so that the distances corresponding to the cropped local geodesic regions are less than or equal to the preset distance threshold.

14. A quality evaluation device for a three-dimensional texture grid, characterized in that: include: A mesh acquisition module, used to acquire a distorted mesh to be evaluated and a reference mesh corresponding to the distorted mesh, wherein the reference mesh and the distorted mesh are three-dimensional meshes corresponding to a texture map; A sampling module, used for sampling points in at least one to-be-processed grid to obtain a target point set, wherein the at least one to-be-processed grid is at least one of the reference grid and the distorted grid; A neighbor point determination module, configured to determine, for each target point in the target point set, a first nearest neighbor point of the target point from the reference grid based on distances between points, and determine a second nearest neighbor point of the target point from the distorted grid; A local area determination module is used for, for each of the target points, taking the first nearest point of the target point as the area center point, determining a first local geodesic area corresponding to the target point from the reference grid, and taking the second nearest point of the target point as the area center point, determining a second local geodesic area corresponding to the target point from the distorted grid; A similarity determination module, configured to determine, for each target point, a similarity between a first local geodesic region and a second local geodesic region corresponding to the target point; The quality evaluation module is used to determine the quality evaluation result of the distorted grid based on the similarity corresponding to each of the target points.

15. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 13.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 13 are implemented.

Images