Point cloud data quality assessment method, device, computer device and storage medium

Abstract

The embodiments of the present application belong to the technical field of data analysis, and relate to a method, device, computer device, and storage medium for evaluating the quality of point cloud data. The present application analyzes the two-dimensional point cloud data collected by traffic sensors, and extracts geometric features and Gaussian projection features to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data, making up for the deficiencies in the point cloud quality evaluation method.

Description

Technical Field

[0001] The present application relates to the technical field of data analysis, and in particular, to a method, device, computer device, and storage medium for evaluating the quality of point cloud data. Background Art

[0002] With the development of intelligent transportation and driverless driving, like image data, point cloud data has gradually become a basic type of data in the field of traffic target detection. The traffic target detection technology based on two-dimensional point cloud data has been widely applied in the field of traffic detection. The task of traffic target detection is to find all objects of interest in the two-dimensional point cloud data, such as vehicles, pedestrians, etc., and mark their positions and categories. At present, traffic target detection mainly extracts the targets in the two-dimensional point cloud data, and then matches the targets with the existing models in the model library to complete the detection and recognition operation of the targets. Therefore, the quality of two-dimensional point cloud data is the key to determining the efficiency and accuracy of traffic target detection.

[0003] However, due to the inherent non-structural and disorderly characteristics of point cloud data, the traditional image quality evaluation methods cannot be directly applied to the field of evaluating the quality of two-dimensional point cloud data. The existing point cloud quality evaluation methods mainly include subjective evaluation methods and objective evaluation methods. Among them, the subjective evaluation methods are less frequently used in point cloud quality evaluation due to their disadvantages such as time-consuming, laborious, and being easily affected by personal factors. The objective evaluation methods get rid of the limitation of relying on human subjective judgment and effectively improve the evaluation efficiency, and are often used for evaluating the quality of point cloud data.

[0004] The applicant has found that the research related to objective evaluation methods mainly focuses on visual reconstruction fields such as trajectory reconstruction and model reconstruction, and most evaluations are carried out from single points, point cloud density, and point cloud elevation, which are relatively general and not comprehensive. At present, there is less research on evaluating the quality of two-dimensional point cloud data in traffic target detection. Therefore, there is an urgent need for a feasible, effective, and accurate method or system for evaluating point cloud data. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to propose a method, device, computer device, and storage medium for evaluating the quality of point cloud data, so as to solve the problem that the traditional point cloud data evaluation methods have low feasibility, effectiveness, and accuracy.

[0006] To solve the above technical problems, the embodiments of the present application provide a method for evaluating the quality of point cloud data, which adopts the following technical solutions:

[0007] Obtain the original point cloud data to be evaluated;

[0008] Perform a feature extraction operation on the original point cloud data to obtain a set of point cloud data features;

[0009] Obtain the weight scores of each feature according to the set of point cloud data features;

[0010] Calculate the information loss evaluation index according to the weight scores;

[0011] Perform a quality assessment operation on the original point cloud data according to the information loss evaluation index to obtain a quality assessment result.

[0012] To solve the above technical problems, an embodiment of the present application also provides a point cloud data quality assessment device, which adopts the following technical solutions:

[0013] An original data acquisition module, configured to acquire the original point cloud data to be evaluated;

[0014] A feature extraction module, configured to perform a feature extraction operation on the original point cloud data to obtain a set of point cloud data features;

[0015] A weight score acquisition module, configured to obtain the weight scores of each feature according to the set of point cloud data features;

[0016] An evaluation index calculation module, configured to calculate the information loss evaluation index according to the weight scores;

[0017] A quality assessment module, configured to perform a quality assessment operation on the original point cloud data according to the information loss evaluation index to obtain a quality assessment result.

[0018] To solve the above technical problems, an embodiment of the present application also provides a computer device, which adopts the following technical solutions:

[0019] It includes a memory and a processor. Computer-readable instructions are stored in the memory, and when the processor executes the computer-readable instructions, the steps of the above-mentioned point cloud data quality assessment method are implemented.

[0020] To solve the above technical problems, an embodiment of the present application also provides a computer-readable storage medium, which adopts the following technical solutions:

[0021] Computer-readable instructions are stored on the computer-readable storage medium, and when the computer-readable instructions are executed by a processor, the steps of the above-mentioned point cloud data quality assessment method are implemented.

[0022] Compared with the prior art, the embodiments of the present application mainly have the following beneficial effects:

[0023] The present application provides a method for evaluating the quality of point cloud data, including: obtaining the original point cloud data to be evaluated; performing a feature extraction operation on the original point cloud data to obtain a set of point cloud data features; obtaining the weight scores of each feature according to the set of point cloud data features; calculating an information loss evaluation index according to the weight scores; and performing a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result. The present application analyzes the two-dimensional point cloud data collected by traffic sensors, extracts geometric features and Gaussian projection features to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data and make up for the deficiencies in the point cloud quality evaluation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the solutions in the present application, the following will briefly introduce the drawings required for the description of the embodiments of the present application. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0025] Figure 1 is an exemplary system architecture diagram to which the present application can be applied;

[0026] Figure 2 is a flowchart of the implementation of the method for evaluating the quality of point cloud data provided in Embodiment 1 of the present application;

[0027] Figure 3 is Figure 2 a flowchart of a specific implementation manner of step S202 in

[0028] Figure 4 is Figure 2 a flowchart of a specific implementation manner of step S203 in

[0029] Figure 5 is Figure 2 a flowchart of a specific implementation manner of step S204 in

[0030] Figure 6 is a flowchart of a specific implementation manner of the method for obtaining target data samples provided in Embodiment 1 of the present application;

[0031] Figure 7 is a schematic diagram of a specific implementation manner of the original point cloud data provided in Embodiment 1 of the present application;

[0032] Figure 8 It is a schematic structural diagram of the point cloud data quality evaluation device provided in the second embodiment of the present application;

[0033] Figure 9 is Figure 8 a schematic structural diagram of a specific implementation manner in

[0034] Figure 10 is Figure 8 a schematic structural diagram of a specific implementation manner in

[0035] Figure 11 It is a schematic structural diagram of an embodiment of a computer device according to the present application. Specific implementation manner

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs; the terms used in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion. The terms "first", "second", etc. in the specification and claims of this application or the above drawings are used to distinguish different objects and not to describe a specific order.

[0037] Referring to "embodiment" herein means that a specific feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of this application. The phrase appears at various positions in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0038] In order to enable those skilled in the technical field to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0039] As Figure 1 shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0040] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0041] Terminal devices 101, 102, and 103 can be various electronic devices with a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV) players, laptop computers, desktop computers, and so on.

[0042] Server 105 can be a server providing various services, such as a background server supporting the pages displayed on terminal devices 101, 102, and 103.

[0043] It should be noted that the point cloud data quality evaluation method provided by the embodiments of the present application is generally executed by the server / terminal device. Correspondingly, the point cloud data quality evaluation device is generally set in the server / terminal device.

[0044] It should be understood that Figure 1 the numbers of the terminal devices, networks, and servers in

[0045] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and servers. Figure 2 Continuing to refer to

[0046] The above-mentioned point cloud data quality evaluation method includes the following steps:

[0047] Step S201: Obtain the original point cloud data to be evaluated.

[0048] In the embodiments of the present application, obtaining the original point cloud data can be to collect two-dimensional point cloud data through traffic sensors, convert the collected data frame by frame into point cloud data and store it in a text file. Each text file is a collection of a series of point cloud data, denoted as {(x i , y i )|i = 0, 1,..., n}, where n is the number of point cloud data.

[0049] Step S202: Perform a feature extraction operation on the original point cloud data to obtain a point cloud data feature set.

[0050] In the embodiment of the present application, the feature extraction operation may be to obtain the minimum circumscribed rectangle of the original point cloud data according to the target edge recognition method, where the minimum circumscribed rectangle carries length data and width data; use the ratio of the length data and the width data as the geometric feature value; calculate the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data in the horizontal direction according to the Gaussian kernel function; calculate the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density of the original point cloud data in the vertical direction according to the Gaussian kernel function; integrate the geometric feature value, the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, the standard deviation of the horizontal probability density, the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density to obtain the point cloud data feature set.

[0051] Step S203: Obtain the weight scores of each feature according to the point cloud data feature set.

[0052] In the embodiment of the present application, the way to obtain the weight scores may be to obtain c groups of point cloud data feature sets in the point cloud data feature set, calculate the first evaluation index of each group of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where c ∈ N*; perform dimensionality reduction processing on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes the first evaluation index, the linear combination coefficient, and the new variable; obtain the covariance matrix corresponding to the new variable feature matrix; perform a QR decomposition operation on the covariance matrix to obtain the covariance eigenvalues; calculate the contribution rate according to the covariance eigenvalues; select the target new variable that meets the weight selection condition in the new variable; calculate the linear combination coefficient of the first evaluation index in the target new variable; calculate the coefficient of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficient of the first evaluation index in the target new variable, and normalize the coefficient to obtain the weight scores.

[0053] Step S204: Calculate the information loss evaluation index according to the weight scores.

[0054] In the embodiments of the present application, the information loss evaluation index can be obtained by performing a data processing operation on each group of the original point cloud data a times according to the same data processing method to obtain a second evaluation index, and constructing an information entropy feature matrix based on the second evaluation index; calculating the information entropy of a single index based on the information entropy feature matrix; and calculating the information loss evaluation index based on the weight score and the information entropy of the single index.

[0055] Step S205: Perform a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result.

[0056] In the embodiments of the present application, a method for evaluating the quality of point cloud data is provided, including: obtaining the original point cloud data to be evaluated; performing a feature extraction operation on the original point cloud data to obtain a set of point cloud data features; obtaining the weight score of each feature according to the set of point cloud data features; calculating an information loss evaluation index according to the weight score; and performing a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result. Through the analysis of the two-dimensional point cloud data collected by traffic sensors, the present application extracts geometric features and Gaussian projection features to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data and make up for the deficiencies in the point cloud quality evaluation method.

[0057] Continue to refer to Figure 3 which shows Figure 2 a flowchart of a specific implementation manner of step S202 in

[0058] In some optional implementation manners of this embodiment, step S202 specifically includes:

[0059] Step S301: Obtain the minimum circumscribed rectangle of the original point cloud data according to the target edge recognition method, where the minimum circumscribed rectangle carries length data and width data.

[0060] Step S302: Use the ratio of the length data and the width data as the geometric feature value.

[0061] In the embodiments of the present application, the point cloud data is visualized, and the minimum circumscribed rectangle of the target is obtained through the target edge recognition method. The ratio of the length h and the width w of the minimum circumscribed rectangle is denoted as the aspect ratio R a and used as feature one, which is represented by formula (1).

[0062]

[0063] Step S303: Calculate the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data according to the Gaussian kernel function.

[0064] In the embodiment of the present application, taking the projection onto the X-axis coordinate as an example, assuming that there are n pixel points in one frame of point cloud data, then the coordinate set of each pixel point after projection is [x1, x2,..., x n . In this paper, the Gaussian kernel function is selected as the kernel function for kernel density estimation, and the calculation formula of the kernel density curve is expressed by formula (2).

[0065]

[0066] In the formula: n is the number of pixel points; v is the smoothing parameter of the Gaussian kernel function; θ is the bandwidth of the Gaussian kernel function; x i is the projection coordinate of the i-th pixel point, x i ∈[x1, x2,..., x n .

[0067] Under the condition of the kernel density curve calculation formula, the minimum value s x , the maximum value l x , the average value u x , and the standard deviation σ x of the probability density of the point cloud data projected onto the X-axis are respectively denoted as Feature Two, Feature Three, Feature Four, and Feature Five, and are expressed by formulas (3) to (6).

[0068] s x =[f(x i )] min (3)

[0069] l x =[f(x i )] max (4)

[0070]

[0071]

[0072] Step S304: Calculate the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density of the original point cloud data according to the Gaussian kernel function.

[0073] In the embodiment of the present application, similarly, the minimum value s y , the maximum value l y, the average value u y and the standard deviation σ y , which are respectively denoted as Feature Six, Feature Seven, Feature Eight, and Feature Nine, and are represented by formulas (7) to (10).

[0074] s y = [f(y i )] min (7)

[0075] l y = [f(y i )] max (8)

[0076]

[0077]

[0078] Step S305: Integrate the geometric eigenvalue, the minimum value of the horizontal probability density, the maximum value of the horizontal probability density, the average value of the horizontal probability density, the standard deviation of the horizontal probability density, the minimum value of the vertical probability density, the maximum value of the vertical probability density, the average value of the vertical probability density, and the standard deviation of the vertical probability density to obtain the point cloud data feature set.

[0079] In the embodiment of the present application, the nine features form a 9-dimensional feature vector T, which is represented by formula (11).

[0080] T = (R a , s x , l x , u x , σ x , s y , l y , u y , σ y ) (11)

[0081] In the embodiment of the present application, it is assumed that the total number of sample points of the original point cloud data is N, and all features are calculated as a data set after data processing of the samples.

[0082] Continue to refer to Figure 4 , which shows Figure 2 a flowchart of a specific implementation manner of step S203 in

[0083] In some optional implementation manners of this embodiment, step S203 specifically includes:

[0084] Step S401: Obtain c groups of point cloud data feature sets from the point cloud data feature set, calculate the first evaluation index of each group of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where c ∈ N*.

[0085] Step S402: Perform dimensionality reduction on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes a first evaluation index, a linear combination coefficient, and a new variable.

[0086] Step S403: Obtain the covariance matrix corresponding to the new variable feature matrix.

[0087] Step S404: Perform a QR decomposition operation on the covariance matrix to obtain covariance eigenvalues.

[0088] Step S405: Calculate the contribution rate according to the covariance eigenvalues.

[0089] Step S406: Select target new variables that meet the weight selection conditions from the new variables.

[0090] Step S407: Calculate the linear combination coefficient of the first evaluation index in the target new variable.

[0091] Step S408: Calculate the coefficient of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficient of the first evaluation index in the target new variable, and normalize the coefficient to obtain a weight score.

[0092] In the embodiment of the present application, c groups of original point cloud data are selected, and n evaluation indexes of each group of data are calculated and standardized to form a matrix D with c rows and n columns. c×n 。

[0093]

[0094] Denote the original evaluation indexes as d1, d2,..., d n , perform dimensionality reduction on them, and denote the transformed new variables as y1, y2,..., y m Expressed as a linear combination of the original variables.

[0095]

[0096] In the formula, l ij is the linear combination coefficient of the jth evaluation index in the new variable y i (i = 1, 2..., m).

[0097] In the embodiment of the present application, the covariance matrix S of the matrix D c×n is obtained by using formulas (14) and (15), and the QR decomposition is performed on the covariance matrix S to obtain the eigenvalues of the covariance matrix, denoted as λ1 ≥ λ2 ≥... ≥ λ n 。

[0098] S = E[(D - E(D))(D - E(D)) T (14)

[0099] P T SP = Diag(λ1, λ2,..., λ n ) (15)

[0100] where P = (e1, e2,... e n ) and e i = (e i1 , e i2 ,... e in ) T , e ij represents the load number of the new variable y i on the evaluation index d j .

[0101] In the embodiment of the present application, the contribution rate r i of each new variable is calculated using formula (16). According to the contribution rate r i , m new variables are selected to determine the weights of each evaluation index in the information entropy calculation. The selection formula is

[0102]

[0103] The linear combination coefficient l ij of the original n evaluation indexes in the m new variables is calculated, and the calculation method is as shown in formula (17):

[0104]

[0105] In the embodiment of the present application, the coefficient x j of each evaluation index in the comprehensive model is calculated using formula (18). The coefficients x j of each evaluation index in the comprehensive model are normalized to obtain the weights w j of the n evaluation indexes, which is represented by formula (19).

[0106]

[0107]

[0108] Continue to refer to Figure 5 , which shows Figure 2 a flowchart of a specific implementation manner of step S204 in

[0109] In some alternative implementation manners of this embodiment, step S204 specifically includes:

[0110] Step S501: Perform a data processing operation on each group of original point cloud data according to the same data processing method to obtain a second evaluation index, and construct an information entropy feature matrix based on the second evaluation index.

[0111] Step S502: Calculate the single-index information entropy according to the information entropy feature matrix.

[0112] Step S503: Calculate the information loss evaluation index according to the weight score and the single-index information entropy.

[0113] Perform a data processing on each group of original point cloud data a times under the same data processing method, calculate n evaluation indexes after the data processing, and standardize them to form a matrix B with a rows and n columns.

[0114] In the embodiment of the present application, calculate the proportion P of the value of the jth evaluation index at the ith data processing to the total number of this index ij , which is represented by formula (20). According to the proportion P ij calculate the entropy value u of the jth index j , and the calculation method is as shown in formula (21).

[0115]

[0116]

[0117] In the formula, k > 0 and is only related to the number of times of processing the point cloud data. Generally, let k = 1 / ln(a).

[0118] In the embodiment of the present application, multiply the information entropy value of a single index by the weight and sum them up to obtain the information loss evaluation index U j .

[0119]

[0120] Continue to refer to Figure 6 , which shows a flowchart of a specific implementation manner of the target data sample acquisition method provided in the first embodiment of the present application. For the convenience of description, only the parts related to the present application are shown.

[0121] In some optional implementation manners of this embodiment, after step S205, it further includes:

[0122] Step S601: Confirm the best data processing method according to the quality evaluation result;

[0123] Step S602: Perform a data processing operation on the original point cloud data according to the best data processing method to obtain a target data sample with a small sample size and little effective information loss.

[0124] In the embodiments of the present application, the information loss evaluation index of all processed point cloud data is calculated through the above steps, and based on this, the quality of the point cloud data is analyzed and evaluated, so as to select the best data processing method. The original data is processed by the selected processing method to obtain a data sample with a small sample size and less loss of effective information.

[0125] In the embodiments of the present application, the sampling method is a commonly used method for processing point cloud data. Therefore, the present invention uses a dynamic vision sensor to collect the original point cloud data, and selects three different sampling methods, namely random sampling, grid sampling, and density sampling, for data processing. The information loss evaluation index is calculated by using the above proposed evaluation method to evaluate the sampled samples, and the accuracy of the evaluation method is verified by the signal-to-noise ratio. The original point cloud data collected by the dynamic vision sensor is as Figure 7 shown, and the statistical results of the signal-to-noise ratio and its loss rate after sampling are shown in Table 1.

[0126] Table 1 Statistical results of signal-to-noise ratio and information loss rate of different sampling methods

[0127]

[0128] In the embodiments of the present application, as can be seen from Table 1, as the sampling rate increases, the change trends of the signal-to-noise ratio and the information loss rate are the same, and the signal-to-noise ratio and the information loss rate of density sampling are better than those of the other two sampling methods. Therefore, the quality of the samples extracted by density sampling is higher, the distribution characteristics of the original data set are effectively maintained, and the anti-noise ability is stronger. For density sampling, when the sampling rate is between 40% and 70%, the extracted vehicles are relatively complete and the number of noise points is small, which is consistent with the results obtained by the evaluation method proposed in the present invention, proving that the evaluation method proposed in the present invention can effectively evaluate the quality of point cloud data.

[0129] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the computer-readable instructions are executed, they can include the processes of the embodiments of the above methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), etc., or a random access memory (RAM), etc.

[0130] It should be understood that although the steps in the flowchart of the accompanying drawings are shown sequentially according to the indication of the arrows, these steps are not necessarily executed sequentially in the order indicated by the arrows. Unless otherwise clearly stated in this document, the execution of these steps has no strict order restriction and can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or sub-steps or stages of other steps.

[0131] Embodiment 2

[0132] Further referring to Figure 8 , as an implementation of the method shown above Figure 2 , this application provides an embodiment of a point cloud data quality evaluation device. This device embodiment corresponds to Figure 2 the method embodiment shown, and this device can be specifically applied to various electronic devices.

[0133] As shown in Figure 8 , the point cloud data quality evaluation device 200 described in this embodiment includes: a raw data acquisition module 210, a feature extraction module 220, a weight score acquisition module 230, an evaluation index calculation module 240, and a quality evaluation module 250. Among them:

[0134] The raw data acquisition module 210 is used to acquire the raw point cloud data to be evaluated;

[0135] The feature extraction module 220 is used to perform feature extraction operations on the raw point cloud data to obtain a point cloud data feature set;

[0136] The weight score acquisition module 230 is used to obtain the weight scores of each feature according to the point cloud data feature set;

[0137] The evaluation index calculation module 240 is used to calculate the information loss evaluation index according to the weight scores;

[0138] The quality evaluation module 250 is used to perform a quality evaluation operation on the raw point cloud data according to the information loss evaluation index to obtain a quality evaluation result.

[0139] In the embodiment of this application, obtaining the raw point cloud data may be to collect two-dimensional point cloud data through a traffic sensor, convert the collected data frame by frame into point cloud data and store it in a text file. Each text file is a set of a series of point cloud data, denoted as {(x i ,y i)|i = 0, 1, ..., n}, where n is the number of point cloud data.

[0140] In an embodiment of the present application, the feature extraction operation may be to obtain the minimum bounding rectangle of the original point cloud data according to the target edge recognition method, where the minimum bounding rectangle carries length data and width data; use the ratio of the length data and the width data as the geometric feature value; calculate the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data in the horizontal direction according to the Gaussian kernel function; calculate the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density of the original point cloud data in the vertical direction according to the Gaussian kernel function; integrate the geometric feature value, the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, the standard deviation of the horizontal probability density, the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density to obtain the point cloud data feature set.

[0141] In an embodiment of the present application, the way to obtain the weight score may be to obtain c groups of point cloud data feature sets in the point cloud data feature set, calculate the first evaluation index of each group of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where c ∈ N*; perform dimensionality reduction processing on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes the first evaluation index, the linear combination coefficient, and the new variable; obtain the covariance matrix corresponding to the new variable feature matrix; perform QR decomposition operation on the covariance matrix to obtain the covariance eigenvalue; calculate the contribution rate according to the covariance eigenvalue; select the target new variable that meets the weight selection condition among the new variables; calculate the linear combination coefficient of the first evaluation index in the target new variable; calculate the coefficient of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficient of the first evaluation index in the target new variable, and normalize the coefficient to obtain the weight score.

[0142] In an embodiment of the present application, calculating the information loss evaluation index may be to perform a data processing operation on each group of the original point cloud data a times according to the same data processing method to obtain a second evaluation index, and construct an information entropy feature matrix according to the second evaluation index; calculate the single-index information entropy according to the information entropy feature matrix; calculate the information loss evaluation index according to the weight score and the single-index information entropy.

[0143] In an embodiment of the present application, a point cloud data quality evaluation device 200 is provided, including: an original data acquisition module 210 for acquiring original point cloud data to be evaluated; a feature extraction module 220 for performing feature extraction operations on the original point cloud data to obtain a point cloud data feature set; a weight score acquisition module 230 for obtaining weight scores of each feature according to the point cloud data feature set; an evaluation index calculation module 240 for calculating an information loss evaluation index according to the weight scores; and a quality evaluation module 250 for performing a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result. In the present application, by analyzing the two-dimensional point cloud data collected by traffic sensors, geometric features and Gaussian projection features are extracted to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data and make up for the deficiencies in the point cloud quality evaluation method.

[0144] Continuing to refer to Figure 9 , shows Figure 8 a schematic structural diagram of a specific implementation manner of the feature extraction module 220 in

[0145] In some optional implementation manners of this embodiment, the above-mentioned feature extraction module 220 includes: a circumscribed rectangle acquisition sub-module 221, a geometric feature value confirmation sub-module 222, a horizontal probability density sub-module 223, a vertical probability density sub-module 224, and a feature integration sub-module 225, where:

[0146] The circumscribed rectangle acquisition sub-module 221 is configured to obtain the minimum circumscribed rectangle of the original point cloud data according to a target edge recognition method, where the minimum circumscribed rectangle carries length data and width data;

[0147] The geometric feature value confirmation sub-module 222 is configured to use the ratio of the length data and the width data as the geometric feature value;

[0148] The horizontal probability density sub-module 223 is configured to calculate the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data in the horizontal direction according to a Gaussian kernel function;

[0149] A vertical probability density sub-module 224, configured to calculate a minimum vertical probability density, a maximum vertical probability density, an average vertical probability density, and a standard deviation of the vertical probability density of the original point cloud data in the vertical direction according to the Gaussian kernel function;

[0150] A feature integration sub-module 225, configured to integrate the geometric feature values, the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, the standard deviation of the horizontal probability density, the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density to obtain the point cloud data feature set.

[0151] In an embodiment of the present application, the point cloud data is visualized, and the minimum circumscribed rectangle of the target is obtained by means of target edge recognition. The ratio of the length h to the width w of the minimum circumscribed rectangle is denoted as the aspect ratio R a And is used as Feature 1, which is represented by formula (1).

[0152]

[0153] In an embodiment of the present application, taking the projection onto the X-axis coordinate as an example, assuming that there are n pixel points in one frame of point cloud data, then the coordinate set of each pixel point after projection is [x1, x2,..., x n . In this paper, the Gaussian kernel function is selected as the kernel function for kernel density estimation, and the kernel density curve calculation formula is represented by formula (2).

[0154]

[0155] Where: n is the number of pixel points; v is the smoothing parameter of the Gaussian kernel function; θ is the bandwidth of the Gaussian kernel function; x i is the projection coordinate of the i-th pixel point, x i ∈[x1, x2,..., x n .

[0156] Under the condition of the kernel density curve calculation formula, the minimum value s x , the maximum value l x , the average value u x , and the standard deviation σ x of the probability density of the point cloud data projected onto the X-axis are respectively denoted as Feature 2, Feature 3, Feature 4, and Feature 5, and are represented by formulas (3) to (6).

[0157] s x =[f(x i )] min (3)

[0158] l x =[f(xi )] max (4)

[0159]

[0160]

[0161] In the embodiment of the present application, similarly, the minimum value s of the kernel density curve and probability density of the point cloud data projected on the Y-axis can be obtained. y , the maximum value l y , the average value u y and the standard deviation σ y , which are respectively denoted as Feature Six, Feature Seven, Feature Eight, and Feature Nine, and are represented by formulas (7) to (10).

[0162] s y =[f(y i )] min (7)

[0163] l y =[f(y i )] max (8)

[0164]

[0165]

[0166] In the embodiment of the present application, the nine features constitute a 9-dimensional feature vector T, which is represented by formula (11).

[0167] T=(R a , s x , l x , u x , σ x , s y , l y , u y , σ y ) (11)

[0168] In the embodiment of the present application, it is assumed that the total number of sample points of the original point cloud data is N, and after data processing of the samples, all features are calculated as a data set.

[0169] Continue to refer to Figure 10 , which shows Figure 8 a schematic structural diagram of a specific implementation manner of the weight score acquisition module 230 in

[0170] In some alternative implementation manners of this embodiment, the above weight score acquisition module 230 includes: a weight feature matrix construction sub-module 231, a dimensionality reduction processing module 232, a covariance matrix acquisition sub-module 233, a QR decomposition sub-module 234, a contribution rate calculation sub-module 235, a target new variable confirmation sub-module 236, a linear combination coefficient calculation sub-module 237, and a weight score confirmation sub-module 238, where:

[0171] The weight feature matrix construction sub-module 231 is configured to obtain c groups of point cloud data feature sets in the point cloud data feature set, calculate a first evaluation index for each group of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where c ∈ N*;

[0172] The dimensionality reduction processing module 232 is configured to perform dimensionality reduction processing on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes the first evaluation index, a linear combination coefficient, and a new variable;

[0173] The covariance matrix acquisition sub-module 233 is configured to obtain a covariance matrix corresponding to the new variable feature matrix;

[0174] The QR decomposition sub-module 234 is configured to perform a QR decomposition operation on the covariance matrix to obtain covariance eigenvalues;

[0175] The contribution rate calculation sub-module 235 is configured to calculate a contribution rate according to the covariance eigenvalues;

[0176] The target new variable confirmation sub-module 236 is configured to select target new variables that meet the weight selection conditions from the new variables;

[0177] The linear combination coefficient calculation sub-module 237 is configured to calculate a linear combination coefficient of the first evaluation index in the target new variable;

[0178] The weight score confirmation sub-module 238 is configured to calculate a coefficient of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficient of the first evaluation index in the target new variable, and normalize the coefficient to obtain the weight score.

[0179] In the embodiment of the present application, c groups of original point cloud data are selected, n evaluation indexes of each group of data are calculated respectively and subjected to standardization processing to form a matrix D with c rows and n columns c×n .

[0180]

[0181] Denote the original evaluation indexes as d1, d2,..., d n, perform dimensionality reduction on it, and denote the transformed new variables as y1, y2,..., y m is expressed as a linear combination of the original variables.

[0182]

[0183] In the formula, l ij is the linear combination coefficient of the jth evaluation index in the new variable y i (i = 1, 2..., m).

[0184] In the embodiment of the present application, the covariance matrix S of the matrix D is obtained by using formulas (14) and (15), and the QR decomposition is performed on the covariance matrix S to obtain the eigenvalues of the covariance matrix, denoted as λ1 ≥ λ2 ≥... ≥ λ c×n . n .

[0185] S = E[(D - E(D))(D - E(D)) T (14)

[0186] P T SP = Diag(λ1, λ2,..., λ n ) (15)

[0187] In the formula, P = (e1, e2,... e n ) and e i = (e i1 , e i2 ,... e in ) T , e ij represents the loading number of the new variable y i for the evaluation index d j .

[0188] In the embodiment of the present application, the contribution rate r i of each new variable is calculated by using formula (16). According to the contribution rate r i , m new variables are selected to determine the weights of each evaluation index in the information entropy calculation. The selection formula is

[0189]

[0190] Calculate the linear combination coefficient l ij of the original n evaluation indexes in the m new variables. The calculation method is as shown in formula (17):

[0191]

[0192] In the embodiment of the present application, the coefficient x of each evaluation index in the comprehensive model is calculated by using formula (18)j For the coefficients x of each evaluation index in the comprehensive model j perform normalization to obtain the weights w of n evaluation indices j , which is represented by formula (19).

[0193]

[0194]

[0195] In some alternative implementation manners of this embodiment, the above point cloud data quality evaluation device 200 further includes: an optimal method confirmation module and a data processing module, where:

[0196] The optimal method confirmation module is configured to confirm the optimal data processing method according to the quality evaluation result;

[0197] The data processing module is configured to perform data processing operations on the original point cloud data according to the optimal data processing method to obtain a target data sample with a small sample size and less loss of effective information.

[0198] In the embodiments of this application, by calculating the information loss evaluation index of all processed point cloud data through the above steps, the quality of the point cloud data is analyzed and evaluated based on this, so as to select the optimal data processing method. The original data is processed through the selected processing method to obtain a data sample with a small sample size and less loss of effective information.

[0199] In the embodiments of this application, the sampling method is a commonly used point cloud data processing method. Therefore, the present invention uses a dynamic vision sensor to collect the original point cloud data, and selects three different sampling methods, namely random sampling, grid sampling, and density sampling, for data processing. The information loss evaluation index is calculated by using the above proposed evaluation method to evaluate the sampled samples, and the signal-to-noise ratio is used to verify the accuracy of the evaluation method. The dynamic vision sensor collects the original point cloud data as Figure 7 shown, and the statistical results of the signal-to-noise ratio and its loss rate after sampling are shown in Table 1.

[0200] Table 1 Statistical results of signal-to-noise ratio and information loss rate of different sampling methods

[0201]

[0202] In the embodiments of the present application, as can be seen from Table 1, as the sampling rate increases, the change trends of the signal-to-noise ratio and the information loss rate are consistent, and both the signal-to-noise ratio and the information loss rate of density sampling are better than those of the other two sampling methods. Therefore, the quality of the samples extracted by density sampling is higher, the distribution characteristics of the original data set are effectively maintained, and the anti-noise ability is stronger. For density sampling, when the sampling rate is between 40% and 70%, the extracted vehicles are relatively complete and the number of noise points is small, which is consistent with the results obtained by the evaluation method proposed in the present invention, proving that the evaluation method proposed in the present invention can effectively evaluate the quality of point cloud data.

[0203] To solve the above technical problems, the embodiments of the present application also provide a computer device. For details, please refer to Figure 11 , Figure 11 which is the basic structural block diagram of the computer device in this embodiment.

[0204] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are communicatively connected to each other through a system bus. It should be noted that only the computer device 300 with components 310-330 is shown in the figure, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Among them, those skilled in the art of the present technology can understand that a computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0205] The computer device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The computer device can perform human-computer interaction with the user through a keyboard, a mouse, a remote control, a touchpad, a voice control device, etc.

[0206] The memory 310 includes at least one type of readable storage medium, which includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 310 may be an internal storage unit of the computer device 300, such as the hard disk or memory of the computer device 300. In other embodiments, the memory 310 may also be an external storage device of the computer device 300, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. equipped on the computer device 300. Of course, the memory 310 may also include both the internal storage unit and the external storage device of the computer device 300. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions of the point cloud data quality assessment method. In addition, the memory 310 may also be used to temporarily store various types of data that have been output or will be output.

[0207] In some embodiments, the processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 320 is generally used to control the overall operation of the computer device 300. In this embodiment, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the point cloud data quality assessment method.

[0208] The network interface 330 may include a wireless network interface or a wired network interface, and this network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0209] The computer device provided in this application analyzes the two-dimensional point cloud data collected by traffic sensors, extracts geometric features and Gaussian projection features to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data and make up for the deficiencies in the point cloud quality assessment method.

[0210] The present application also provides another implementation manner, that is, to provide a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor, so that the at least one processor executes the steps of the point cloud data quality evaluation method as described above.

[0211] The computer-readable storage medium provided by the present application analyzes the two-dimensional point cloud data collected by traffic sensors, extracts geometric features and Gaussian projection features to describe the target. The geometric feature is mainly the aspect ratio of the circumscribed rectangle of the target point cloud distribution, and the Gaussian projection feature includes the horizontal and vertical projection features of the point cloud distribution. Based on the obtained geometric features and projection features, the weights of each feature are calculated, and an evaluation model for the loss degree of point cloud data based on information entropy is established, which can comprehensively and effectively evaluate the quality of two-dimensional point cloud data and make up for the deficiencies in the point cloud quality evaluation method.

[0212] Through the description of the above implementation manners, those skilled in the art can clearly understand that the above-described embodiment method can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation manner. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disc), and includes several instructions for causing a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in various embodiments of the present application.

[0213] Obviously, the above-described embodiments are only part of the embodiments of the present application, rather than all embodiments. The accompanying drawings show preferred embodiments of the present application, but do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosed content of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions described in the foregoing specific implementation manners, or perform equivalent replacements on some of the technical features. Any equivalent structure directly or indirectly using the content of the specification and drawings of the present application in other related technical fields shall be within the scope of the patent protection of the present application by the same token.

Claims

1. A method for evaluating the quality of point cloud data, characterized in that, Including the following steps: Obtain the original point cloud data to be evaluated; Perform feature extraction operations on the original point cloud data to obtain a set of point cloud data features; Obtain the weight scores of each feature according to the set of point cloud data features; Calculate the information loss evaluation index according to the weight scores; Perform a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result; The step of performing feature extraction operations on the original point cloud data to obtain a set of point cloud data features specifically includes the following steps: Obtain the minimum circumscribed rectangle of the original point cloud data according to the target edge recognition method, where the minimum circumscribed rectangle carries length data and width data; Take the ratio of the length data and the width data as the geometric feature value; Calculate the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data in the horizontal direction according to the Gaussian kernel function; Calculate the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density of the original point cloud data in the vertical direction according to the Gaussian kernel function; Integrate the geometric feature value, the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, the standard deviation of the horizontal probability density, the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density to obtain the set of point cloud data features.

2. The method for evaluating the quality of point cloud data according to claim 1, characterized in that, The step of obtaining the weight scores of each feature according to the set of point cloud data features specifically includes the following steps: Obtain c sets of point cloud data feature sets from the point cloud data feature set, calculate the first evaluation index of each set of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where the ; Perform dimensionality reduction processing on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes the first evaluation index, the linear combination coefficient, and the new variable; Obtain the covariance matrix corresponding to the new variable feature matrix; Perform a QR decomposition operation on the covariance matrix to obtain covariance eigenvalues; Calculate the contribution rate according to the covariance eigenvalues; Select target new variables that meet the weight selection conditions among the new variables; Calculate the linear combination coefficient of the first evaluation index in the target new variables; Calculate the coefficient of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficient of the first evaluation index in the target new variables, and normalize the coefficients to obtain the weight scores.

3. The method for evaluating the quality of point cloud data according to claim 1, characterized in that, The step of calculating the information loss evaluation index according to the weight scores specifically includes the following steps: Perform a data processing operations on each group of the original point cloud data according to the same data processing method to obtain a second evaluation index, and construct an information entropy feature matrix based on the second evaluation index, where the ; Calculate the single-index information entropy according to the information entropy feature matrix; Calculate the information loss evaluation index according to the weight scores and the single-index information entropy.

4. The method for evaluating the quality of point cloud data according to claim 1, characterized in that, After the step of performing a quality evaluation operation on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result, the following steps are further included: Confirm the best data processing method according to the quality evaluation result; Perform data processing operations on the original point cloud data according to the optimal data processing method to obtain a target data sample with a small sample size and less loss of effective information.

5. A device for evaluating the quality of point cloud data, characterized in that, Including: An original data acquisition module for acquiring the original point cloud data to be evaluated; A feature extraction module for performing feature extraction operations on the original point cloud data to obtain a set of point cloud data features; A weight score acquisition module for obtaining the weight scores of each feature according to the set of point cloud data features; An evaluation index calculation module for calculating an information loss evaluation index according to the weight scores; A quality evaluation module for performing quality evaluation operations on the original point cloud data according to the information loss evaluation index to obtain a quality evaluation result; The feature extraction module includes: A minimum bounding rectangle acquisition sub-module for obtaining the minimum bounding rectangle of the original point cloud data according to a target edge recognition method, where the minimum bounding rectangle carries length data and width data; A geometric feature value confirmation sub-module for using the ratio of the length data and the width data as the geometric feature value; A horizontal probability density sub-module for calculating the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, and the standard deviation of the horizontal probability density of the original point cloud data in the horizontal direction according to a Gaussian kernel function; A vertical probability density sub-module for calculating the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density of the original point cloud data in the vertical direction according to the Gaussian kernel function; A feature integration sub-module for integrating the geometric feature value, the minimum horizontal probability density, the maximum horizontal probability density, the average horizontal probability density, the standard deviation of the horizontal probability density, the minimum vertical probability density, the maximum vertical probability density, the average vertical probability density, and the standard deviation of the vertical probability density to obtain the set of point cloud data features.

6. The point cloud data quality evaluation device according to claim 5, characterized in that, The weight score acquisition module includes: The weight feature matrix construction sub-module is used to obtain c groups of point cloud data feature sets from the point cloud data feature set, calculate the first evaluation index of each group of point cloud data respectively, and construct a weight feature matrix according to the first evaluation index, where the ; A dimensionality reduction processing module for performing dimensionality reduction processing on the weight feature matrix to obtain a new variable feature matrix, where the new variable feature matrix includes the first evaluation index, linear combination coefficients, and new variables; A covariance matrix acquisition sub-module for obtaining a covariance matrix corresponding to the new variable feature matrix; A QR decomposition sub-module for performing QR decomposition operations on the covariance matrix to obtain covariance eigenvalues; A contribution rate calculation sub-module for calculating the contribution rate according to the covariance eigenvalues; A target new variable confirmation sub-module for selecting target new variables that meet the weight selection conditions from the new variables; A linear combination coefficient calculation sub-module for calculating the linear combination coefficients of the first evaluation index in the target new variables; A weight score confirmation sub-module for calculating the coefficients of each first evaluation index in the comprehensive model according to the contribution rate and the linear combination coefficients of the first evaluation index in the target new variables, and normalizing the coefficients to obtain the weight scores.

7. A computer device, characterized in that, It includes a memory and a processor. Computer-readable instructions are stored in the memory. When the processor executes the computer-readable instructions, the steps of the point cloud data quality assessment method described in any one of claims 1 to 4 are implemented.

8. A computer-readable storage medium, characterized in that, Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are executed by the processor, the steps of the point cloud data quality assessment method described in any one of claims 1 to 4 are implemented.

Images