Method and product for predicting geometric just noticeable distortion of three-dimensional point cloud

Abstract

The invention provides a geometric just-noticeable distortion prediction method and product of a three-dimensional point cloud, and the method comprises the steps: extracting geometric features based on a reference point cloud and a corresponding distortion point cloud, and extracting texture features based on the reference point cloud and the corresponding distortion point cloud; carrying out minimum and maximum normalization processing on the geometric features and the texture features and then splicing the geometric features and the texture features into feature vectors; and inputting the feature vector into a multi-layer perceptron, and outputting a point cloud visual quality prediction result. Through the implementation of the scheme of the invention, the geometric features and the texture features of the three-dimensional point cloud are directly extracted, and the visual quality prediction of the three-dimensional point cloud is carried out on the spliced six-dimensional features based on the multilayer perceptron, so that the point cloud information is fully utilized, and the accuracy of the visual quality prediction result of the three-dimensional point cloud is effectively improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method and product for predicting geometrically just perceptible distortion of a three-dimensional point cloud. Background Art

[0002] Three-dimensional point cloud is a commonly used data form for virtual reality and augmented reality. It contains both three-dimensional coordinates and color information, and can express the objective world more accurately and completely. In practical applications, three-dimensional point clouds usually need to be encoded, and the core of the encoding task is to provide visual quality equivalent to that of the original point cloud at a lower bit rate. In related technologies, the commonly used three-dimensional point cloud visual quality prediction scheme is to project the three-dimensional point cloud onto multiple two-dimensional planes for feature extraction, and then fuse these features to evaluate the visual quality of the point cloud. However, information loss is inevitable in the process of point cloud projection, and because only effective viewpoints are considered, the relevant information of the model cannot be fully utilized, resulting in poor accuracy of the visual quality evaluation results of the three-dimensional point cloud.

[0003] It is worth noting that the techniques described in this section are not necessarily techniques that have been previously conceived or employed. Unless otherwise indicated, it should not be assumed that any technique described in this section is considered to be prior art simply because it is included in this section. Similarly, unless otherwise indicated, the issues mentioned in this section should not be considered to have been recognized in any prior art. Summary of the invention

[0004] The main purpose of this application is to provide a method and product for predicting geometrically perceptible distortion of three-dimensional point clouds, which can at least solve at least one of the above problems provided by the relevant technology.

[0005] The first aspect of the present application provides a method for predicting geometrically perceptible distortion of a three-dimensional point cloud, comprising:

[0006] Extracting geometric features based on a reference point cloud and a corresponding distorted point cloud, and extracting texture features based on the reference point cloud and the corresponding distorted point cloud; wherein the geometric features include a mean curvature difference index, a normal vector spacing, and a number of voxels, and the texture features include a color difference index, a kurtosis difference index, and a brightness difference index;

[0007] Performing minimum and maximum normalization processing on the geometric features and the texture features and then splicing them into a feature vector;

[0008] The feature vector is input into a multilayer perceptron to output a point cloud visual quality prediction result; wherein the point cloud visual quality prediction result includes perceptible distortion and imperceptible distortion.

[0009] The second aspect of the present application provides an electronic device, comprising: a memory and a processor, wherein the processor is used to execute a computer program stored in the memory, and when the processor executes the computer program, it implements each step of the method for predicting geometrically perceptible distortion of a three-dimensional point cloud provided in the first aspect of the embodiment of the present application.

[0010] The third aspect of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for predicting geometrically perceptible distortion of a three-dimensional point cloud provided in the first aspect of the embodiment of the present application are implemented.

[0011] As can be seen from the above, according to the geometrically perceptible distortion prediction method and product of the three-dimensional point cloud provided by the present application, geometric features are extracted based on the reference point cloud and the corresponding distorted point cloud, and texture features are extracted based on the reference point cloud and the corresponding distorted point cloud; the geometric features and texture features are processed by minimum and maximum normalization and then spliced ​​into feature vectors; the feature vectors are input into a multi-layer perceptron to output the point cloud visual quality prediction results. Through the implementation of the present application, the geometric features and texture features of the three-dimensional point cloud are directly extracted, and the visual quality of the three-dimensional point cloud is predicted based on the spliced ​​six-dimensional features based on the multi-layer perceptron, which makes full use of the point cloud information and effectively improves the accuracy of the visual quality prediction results of the three-dimensional point cloud.

[0012] It should be understood that the content described in this section is not intended to identify the key or important features of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings exemplarily illustrate the embodiments and constitute a part of the specification, and together with the text description of the specification, are used to explain the exemplary implementation of the embodiments. The drawings shown are for illustrative purposes only and do not limit the scope of the claims. In all drawings, the same reference numerals refer to similar but not necessarily identical elements.

[0014] Figure 1 A schematic diagram of the principle of a method for predicting geometrically just perceptible distortion of a three-dimensional point cloud provided in one embodiment of the present application;

[0015] Figure 2 A schematic diagram of a basic flow chart of a method for predicting geometrically just perceptible distortion of a three-dimensional point cloud provided in one embodiment of the present application;

[0016] Figure 3 A schematic diagram of functional modules of a geometrically perceptible distortion threshold prediction device provided in one embodiment of the present application;

[0017] Figure 4A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0019] In the description of the embodiments of the present application, the terms "first", "second", etc. are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present application, the term "plurality" means two or more, unless otherwise clearly and specifically defined.

[0020] In order to solve the problem that the point cloud visual quality assessment results of the three-dimensional point cloud visual quality prediction solution provided by the related art are less accurate, an embodiment of the present application provides a method for predicting geometrically perceptible distortion of a three-dimensional point cloud, such as Figure 1 The figure shows the principle schematic diagram of the geometric just noticeable distortion prediction method of the three-dimensional point cloud provided by the present embodiment. The present embodiment models the just noticeable distortion perception calculation as a binary classification problem. Given a reference point cloud Pref and a distorted point cloud Pdist, a mapping relationship between the point cloud pair and the just noticeable distortion threshold is established, where 0 represents noticeable distortion (JND) and 1 represents imperceptible distortion. Specifically, the present embodiment first establishes a point-based neighborhood relationship for the reference point cloud and the distorted point cloud, and then extracts features from the two dimensions of geometry and texture for both the reference point cloud and the distorted point cloud. Next, the geometric features and texture features are spliced, and finally the spliced ​​feature vector is input into a multi-layer perceptron for classification and recognition, and the point cloud visual quality prediction result is output. The point cloud visual quality prediction result is used to indicate whether the point cloud visual quality is noticeable distortion or imperceptible distortion. It is worth mentioning that this embodiment establishes a nonlinear mapping relationship between feature vectors and just noticeable distortion thresholds through a multi-layer perceptron, that is, learning through a training set, obtaining the weight of each feature, and learning the mapping relationship between geometric features and corresponding just noticeable distortion thresholds.

[0021] like Figure 2 The basic flow chart of the method for predicting geometrically perceptible distortion of a three-dimensional point cloud provided in this embodiment includes the following steps:

[0022] Step 201: extracting geometric features based on a reference point cloud and a corresponding distorted point cloud, and extracting texture features based on a reference point cloud and a corresponding distorted point cloud.

[0023] Specifically, the geometric features of this embodiment include the average curvature difference index, the normal vector spacing and the number of voxels, and the texture features include: the color difference index, the kurtosis difference index and the brightness difference index. The quality loss of the distorted point cloud is caused by geometric compression, which is mainly manifested as the displacement or loss of three-dimensional points in the distorted point cloud. Since the three-dimensional points themselves carry color attributes, they will also cause distortion in the texture. In order to extract effective features from the point cloud, this embodiment considers extracting features from both geometric and texture aspects.

[0024] Given a 3D point cloud containing N points Each point is a high-dimensional vector consisting of geometric coordinates and attribute information. The i-th point in P is in represents geometric coordinates, Represents color attribute information.

[0025] Originally, the point cloud uses RGB color to represent the color attribute of each point. Because the RGB components are highly correlated and have certain redundancy, in this embodiment, RGB is converted to YUV color space, where Y represents brightness, and U and V represent two chromaticity components. The conversion formula is expressed as:

[0026]

[0027] In this embodiment, the point p can be found in the reference point cloud and the distorted point cloud respectively by using the K nearest neighbor method. i A set of neighboring points The distance between points is expressed as Euclidean distance In this embodiment, k is selected as 12, thereby establishing point neighborhood relationships in the reference point cloud and the distorted point cloud, respectively. In this embodiment, points in the reference point cloud and the distorted point cloud are matched, and a pair of matching points in the two point clouds are corresponding points.

[0028] In some embodiments, the above-mentioned extraction of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: establishing a point-based neighborhood relationship for the reference point cloud and the corresponding distorted point cloud respectively; calculating the curvature of each point for the neighborhood of each point in the reference point cloud and the distorted point cloud respectively; calculating the curvature difference index between each point in the reference point cloud and the corresponding point in the distorted point cloud; averaging the curvature difference indexes of all points to obtain an average curvature difference index f 1 .

[0029] Specifically, in this embodiment, in the reference point cloud and the distorted point cloud, the curvature is obtained based on the neighborhood of each point, and the curvature calculation is fitted by a quadratic surface, and the curvature is defined by the parameters of the quadratic surface as follows:

[0030]

[0031] Among them, a, b, c, d, and e represent the quadratic surface fitting parameters.

[0032] In this embodiment, when calculating the curvature difference index, the calculation formula is as follows:

[0033]

[0034] Among them, ρ p represents the curvature of the point in the reference point cloud, represents the curvature of the point in the reference point cloud, k 1 Represents a constant, which is used to prevent abnormal calculations when the value is close to 0.

[0035] In some embodiments, the above-mentioned extraction of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: calculating the normal vector of each point for the reference point cloud and the corresponding distorted point cloud respectively; calculating the normal vector distance between each point in the reference point cloud and the corresponding point in the distorted point cloud; averaging the normal vector distances of all points to obtain the normal vector distance f of the entire point cloud. 2 .

[0036] Specifically, in this embodiment, the normal vector distance between the corresponding points in the reference point cloud and the distorted point cloud is defined as follows:

[0037]

[0038] Among them, V p represents the normal vector of a point in the reference point cloud, Represents the normal vector of a point in the distorted point cloud, and dot() represents the vector dot product operation.

[0039] In some embodiments, the extraction of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: obtaining a minimum bounding box for the reference point cloud and the corresponding distorted point cloud respectively; based on the minimum bounding box, evenly dividing the reference point cloud and the distorted point cloud into N cubes respectively; calculating the difference between the number of points in each corresponding cube in the reference point cloud and the distorted point cloud respectively, to obtain an N-dimensional vector; counting the number of non-zero values ​​in the N-dimensional vector as the number of voxels f 3 .

[0040] In this embodiment, voxelization is performed on the reference point cloud and the distorted point cloud, respectively. First, the minimum bounding box of the reference point cloud is obtained (x_max, y_max, z_max), and it is divided into m×n×s small cubes from the three directions of x, y, and z. For example, in this embodiment, the values ​​of m, n, and s are 10, that is, the point cloud model is divided into 1000 small grids. Then, the difference between the number of points in each grid in the reference point cloud and the distorted point cloud is calculated, and a 1000-dimensional vector is obtained. Finally, the number of values ​​in the vector that are not 0 is counted to obtain the number of voxels.

[0041] In some embodiments, the above-mentioned extraction of texture features based on the reference point cloud and the corresponding distorted point cloud includes: calculating the color information of each point for the reference point cloud and the corresponding distorted point cloud respectively; calculating the color information difference between each point in the reference point cloud and the corresponding point in the distorted point cloud; averaging the color information differences of all points to obtain the color difference index f of the entire point cloud. 4 .

[0042] In this embodiment, the calculation formula of the color information is expressed as follows:

[0043]

[0044] Among them, σ rg and μ rg Respectively represent the standard deviation and mean of rg, σ by and μ by They represent the standard deviation and mean of by respectively, and R, G, and B represent the color components of all points in the point cloud respectively.

[0045] In some embodiments, the above-mentioned extraction of texture features based on the reference point cloud and the corresponding distorted point cloud includes: establishing a point-based neighborhood relationship for the reference point cloud and the corresponding distorted point cloud respectively; calculating the kurtosis index of each point for the neighborhood of each point in the reference point cloud and the distorted point cloud respectively; averaging the kurtosis index of all points in the reference point cloud to obtain the overall kurtosis index of the reference point cloud, and averaging the kurtosis index of all points in the distorted point cloud to obtain the overall kurtosis index of the distorted point cloud; calculating the difference between the overall kurtosis index of the reference point cloud and the overall kurtosis index of the distorted point cloud to obtain the kurtosis difference index f 5 .

[0046] This embodiment extracts the kurtosis index for each neighborhood to measure the degree of deviation between a point and its neighboring points. i , the kurtosis index calculation formula of its brightness attribute is:

[0047]

[0048] in, For point pi The neighborhood of L contains k points. j is the brightness value of the jth point, Neighborhood The larger the kurtosis value, the greater the deviation of the point set in the neighborhood, and the smaller the kurtosis value, the smaller the deviation.

[0049] For the reference point cloud and the distorted point cloud, the kurtosis index of all points is averaged to obtain the kurtosis index of the entire point cloud. The calculation formula is expressed as:

[0050]

[0051] Next, the difference between the kurtosis index of the reference point cloud and the distorted point cloud is calculated to obtain the kurtosis difference index that measures the difference in neighborhood attribute distribution.

[0052] In some embodiments, the above-mentioned extraction of texture features based on the reference point cloud and the corresponding distorted point cloud includes: establishing a point-based neighborhood relationship for the reference point cloud and the corresponding distorted point cloud respectively; calculating the neighborhood average brightness difference of the corresponding point based on the neighborhood of each corresponding point in the reference point cloud and the distorted point cloud; averaging the neighborhood average brightness differences of all corresponding points to obtain a brightness difference index f 6 .

[0053] Among them, the calculation formula of the neighborhood average brightness difference is expressed as:

[0054]

[0055] Among them, N represents the number of points in the point cloud, k represents the number of points in the neighborhood, and L i,j and They represent the brightness value of the jth point in the neighborhood of the i-th point in the reference point cloud and the distorted point cloud respectively.

[0056] Step 202: Perform minimum and maximum normalization processing on the geometric features and texture features and then concatenate them into a feature vector.

[0057] Specifically, this embodiment performs a min-max normalization process (Min-Max Scaling) on ​​the geometric features and texture features, that is, mapping the data to a specified range through a linear transformation, and performing feature splicing after normalization to obtain a six-dimensional feature.

[0058] Step 203: Input the feature vector into a multi-layer perceptron and output a point cloud visual quality prediction result.

[0059] The point cloud visual quality prediction results of this embodiment include perceptible distortion and imperceptible distortion. In this embodiment, the extracted feature vector is input into a multi-layer perceptron, and the output is a binary classification result. The multi-layer perceptron can be implemented in a variety of configurations. In the preferred implementation of this embodiment, five fully connected layers can be used, the number of neurons in the first layer (input layer) is 6, the number of neurons in the second layer is 512, the number of neurons in the third layer is 1024, the number of neurons in the fourth layer is 512, and the number of neurons in the fifth layer (output layer) is 2. In addition, the hidden layer activation function can be a rectified linear unit (Rectified Linear Units, Relus), and the output layer classification function can use a Softmax function. The prediction results represent perceptible distortion and imperceptible distortion, respectively.

[0060] In this embodiment, the multilayer perceptron can be expressed as follows:

[0061]

[0062] Among them, y n are the prediction categories, which are perceptible distortion and imperceptible distortion, respectively, and K is the number of categories. In this embodiment, K=2. is the output of the kth neuron.

[0063] It should be understood that the size of the serial number of each step in this embodiment does not mean the order of execution of the steps. The execution order of each step should be determined by its function and internal logic, and should not constitute a sole limitation on the implementation process of the embodiment of this application.

[0064] Next, an embodiment of the present application further provides a device for predicting a geometrically perceptible distortion threshold of a three-dimensional point cloud. Figure 3 This is a functional module diagram of a geometrically perceptible distortion threshold prediction device provided in one embodiment of the present application. The geometrically perceptible distortion threshold prediction device can be used to implement the geometrically perceptible distortion prediction method of a three-dimensional point cloud in the aforementioned embodiment, and mainly includes:

[0065] A feature extraction module 301 is used to extract geometric features based on the reference point cloud and the corresponding distorted point cloud, and to extract texture features based on the reference point cloud and the corresponding distorted point cloud; wherein the geometric features include a mean curvature difference index, a normal vector spacing, and a number of voxels, and the texture features include a color difference index, a kurtosis difference index, and a brightness difference index;

[0066] A feature processing module 302 is used to perform minimum and maximum normalization processing on the geometric features and texture features and then splice them into a feature vector;

[0067] The point cloud prediction module 303 is used to input the feature vector into a multi-layer perceptron and output a point cloud visual quality prediction result; wherein the point cloud visual quality prediction result includes perceptible distortion and imperceptible distortion.

[0068] It should be noted that the geometrically just noticeable distortion prediction methods in the aforementioned embodiments can all be implemented based on the geometrically just noticeable distortion threshold prediction device provided in this embodiment. Ordinary technical personnel in the relevant field can clearly understand that, for the convenience and conciseness of description, the specific working process of the geometrically just noticeable distortion threshold prediction device described in this embodiment can be implemented by referring to the corresponding working process in the aforementioned method embodiment, and will not be repeated here.

[0069] Based on the technical solution of the embodiment of the present application, geometric features are extracted based on the reference point cloud and the corresponding distorted point cloud, and texture features are extracted based on the reference point cloud and the corresponding distorted point cloud; the geometric features and texture features are processed by minimum and maximum normalization and then spliced ​​into feature vectors; the feature vectors are input into a multi-layer perceptron to output the point cloud visual quality prediction results. Through the implementation of the solution of the present application, the geometric features and texture features of the three-dimensional point cloud are directly extracted, and the visual quality of the three-dimensional point cloud is predicted based on the spliced ​​six-dimensional features of the multi-layer perceptron, which makes full use of the point cloud information and effectively improves the accuracy of the visual quality prediction results of the three-dimensional point cloud.

[0070] Figure 4 An electronic device is provided for one embodiment of the present application. The electronic device can be used to implement the geometrically just perceptible distortion prediction method of a three-dimensional point cloud in the aforementioned embodiment, and mainly includes: a memory 401 and a processor 402. The memory 401 stores a computer program 403 that can be run on the processor 402. The memory 401 and the processor 402 are in communication connection. When the processor 402 executes the computer program 403, the geometrically just perceptible distortion prediction method of a three-dimensional point cloud in the aforementioned embodiment is implemented. The number of processors 402 can be one or more.

[0071] The memory 401 may be a high-speed random access memory (RAM) memory, or a non-volatile memory, such as a disk memory. The memory 401 is used to store executable program codes, and the processor 402 is coupled to the memory 401 .

[0072] Furthermore, the present application also provides a computer-readable storage medium, which may be provided in the electronic device in the above embodiments. Figure 4 Memory in the illustrated embodiment.

[0073] The computer readable storage medium stores a computer program, and when the computer program is executed by the processor, the method for predicting geometrically perceptible distortion of a three-dimensional point cloud in the aforementioned embodiment is implemented. Furthermore, the computer storable medium may also be a U disk, a mobile hard disk, a read-only memory (ROM), a RAM, a magnetic disk, or an optical disk, etc., which may store program codes.

[0074] It should be understood that the devices and methods disclosed in the embodiments provided in the present application can also be implemented in any other equivalent manner. For example, the device embodiments described above are only schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0075] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0076] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.

[0077] If the integrated module is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a readable storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned readable storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0078] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described action sequence, because according to the present application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0079] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0080] The above is a description of the geometrically just perceptible distortion prediction method and product of the three-dimensional point cloud provided in this application. For technicians in this field, according to the ideas of the embodiments of this application, there may be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on this application.

Claims

1. A method for predicting geometrically perceptible distortion of a three-dimensional point cloud, characterized in that: include: Extracting geometric features based on a reference point cloud and a corresponding distorted point cloud, and extracting texture features based on the reference point cloud and the corresponding distorted point cloud; wherein the geometric features include a mean curvature difference index, a normal vector spacing, and a number of voxels, and the texture features include a color difference index, a kurtosis difference index, and a brightness difference index; Performing minimum and maximum normalization processing on the geometric features and the texture features and then splicing them into a feature vector; The feature vector is input into a multilayer perceptron to output a point cloud visual quality prediction result; wherein the point cloud visual quality prediction result includes perceptible distortion and imperceptible distortion.

2. The method for predicting geometrically perceptible distortion according to claim 1, characterized in that: The extracting of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: A point-based neighborhood relationship is established for the reference point cloud and the corresponding distorted point cloud; Calculating the curvature of each point in the reference point cloud and the neighborhood of each point in the distorted point cloud respectively; Calculating a curvature difference index between each point in the reference point cloud and a corresponding point in the distorted point cloud; The curvature difference index of all points is averaged to obtain an average curvature difference index.

3. The method for predicting geometrically perceptible distortion according to claim 1, characterized in that: The extracting of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: Calculate the normal vector of each point for the reference point cloud and the corresponding distorted point cloud; Calculating the normal vector distance between each point in the reference point cloud and the corresponding point in the distorted point cloud; The normal vector spacing of all points is averaged to obtain the normal vector spacing of the entire point cloud.

4. The method for predicting geometrically perceptible distortion according to claim 1, wherein: The extracting of geometric features based on the reference point cloud and the corresponding distorted point cloud includes: Obtain the minimum bounding box for the reference point cloud and the corresponding distorted point cloud respectively; Based on the minimum bounding box, the reference point cloud and the distorted point cloud are divided into N cubes respectively; Calculate the difference between the number of points in each corresponding cube in the reference point cloud and the distorted point cloud respectively to obtain an N-dimensional vector; The number of non-zero values ​​in the N-dimensional vector is counted as the number of voxels.

5. The method for predicting geometrically perceptible distortion according to claim 1, characterized in that: The extracting of texture features based on the reference point cloud and the corresponding distorted point cloud includes: Calculate the color information of each point for the reference point cloud and the corresponding distorted point cloud respectively; Calculate the color information difference between each point in the reference point cloud and the corresponding point in the distorted point cloud; The color information difference values ​​of all points are averaged to obtain the color difference index of the entire point cloud.

6. The method for predicting geometrically perceptible distortion according to claim 1, characterized in that: The extracting of texture features based on the reference point cloud and the corresponding distorted point cloud includes: A point-based neighborhood relationship is established for the reference point cloud and the corresponding distorted point cloud; Calculating the kurtosis index of each point for the neighborhood of each point in the reference point cloud and the distorted point cloud respectively; Based on the average value of the kurtosis indexes of all points in the reference point cloud, an overall kurtosis index of the reference point cloud is obtained; and based on the average value of the kurtosis indexes of all points in the distorted point cloud, an overall kurtosis index of the distorted point cloud is obtained; The difference between the overall kurtosis index of the reference point cloud and the overall kurtosis index of the distorted point cloud is calculated to obtain a kurtosis difference index.

7. The method for predicting geometrically perceptible distortion according to claim 1, characterized in that: The extracting of texture features based on the reference point cloud and the corresponding distorted point cloud includes: A point-based neighborhood relationship is established for the reference point cloud and the corresponding distorted point cloud; Based on the neighborhood of each corresponding point in the reference point cloud and the distorted point cloud, calculating the average brightness difference of the neighborhood of the corresponding point; The brightness difference index is obtained by averaging the neighborhood average brightness differences of all the corresponding points.

8. The method for predicting geometrically perceptible distortion according to any one of claims 1 to 7, characterized in that: The multilayer perceptron includes five fully connected layers, wherein the number of neurons in the first fully connected layer is 6, the number of neurons in the second fully connected layer is 512, the number of neurons in the third fully connected layer is 1024, the number of neurons in the fourth fully connected layer is 512, and the number of neurons in the fifth fully connected layer is 2.

9. An electronic device, characterized in that: The device comprises a memory and a processor, wherein: The processor is used to execute the computer program stored in the memory; When the processor executes the computer program, the steps of the method for predicting geometrically perceptible distortion of a three-dimensional point cloud as described in any one of claims 1 to 8 are implemented.

10. 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 for predicting geometrically perceptible distortion of a three-dimensional point cloud as described in any one of claims 1 to 8 are implemented.

Images