Point cloud validity testing methods, apparatus, equipment and storage media

Abstract

This invention relates to the field of point cloud localization and discloses a method, apparatus, device, and storage medium for point cloud validity detection. The method includes: acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data; calculating the proportion information of various types of point cloud data in the processed point cloud data; identifying the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculating the deviation value between the initial pose and the optimized pose; and determining the validity of the raw point cloud data based on the proportion information and the deviation value. The technical solution of this application classifies the acquired raw point cloud data, determines several proportion information and deviation values ​​after classification, and identifies the environmental data at the time of raw point cloud data acquisition, thereby quickly determining the validity of the raw point cloud data and facilitating subsequent further processing of the raw point cloud data.

Description

Technical Field

[0001] This invention relates to the field of point cloud localization, and more particularly to a method, apparatus, device, and storage medium for point cloud validity detection. Background Technology

[0002] LiDAR point cloud localization technology is widely used in autonomous driving, robotics, and other fields. Single-frame point cloud data obtained from an onboard LiDAR is matched with a prior high-precision point cloud map. If the environment remains relatively unchanged, the optimal matching point can always be found in the prior high-precision point cloud map; the coordinates of this point are the vehicle's relative coordinates. However, in reality, road environments frequently change, and the obstruction of LiDAR by other large vehicles around the vehicle can affect the usability of the raw point cloud data. Using raw point cloud data without determining its validity can easily lead to severe distortion in subsequent processing results. Summary of the Invention

[0003] The main objective of this invention is to solve the technical problem in existing point cloud validity detection methods that cannot determine the reliability of original point cloud data based on the original point cloud data acquisition situation.

[0004] The first aspect of the present invention provides a point cloud validity detection method, the point cloud validity detection method comprising: acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data; calculating the proportion information of various types of point cloud data in the processed point cloud data; identifying the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculating the deviation value between the initial pose and the optimized pose; and judging the validity of the raw point cloud data based on the proportion information and the deviation value.

[0005] Optionally, in a first implementation of the first aspect of the present invention, the step of acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data includes: performing vehicle removal processing on the raw point cloud data to obtain vehicle-removed point cloud data; performing distortion compensation processing on the vehicle-removed point cloud data to obtain compensated point cloud data; performing rule detection and classification on the compensated point cloud data to obtain point cloud categories for each compensated point cloud data, and labeling the corresponding compensated point cloud data with the point cloud categories to obtain processed point cloud data.

[0006] Optionally, in a second implementation of the first aspect of the present invention, the point cloud category includes ground point cloud and non-ground point cloud; the calculation of the proportion information of each type of point cloud data in the processed point cloud data includes: calculating the proportion of the vehicle-excluding point cloud data to the original point cloud data to obtain the vehicle-excluding proportion value; calculating the proportion of the processed point cloud data whose point cloud category is ground point cloud to the vehicle-excluding point cloud data to obtain the ground proportion value.

[0007] Optionally, in a third implementation of the first aspect of the present invention, the step of calculating the proportion information of various types of point cloud data in the processed point cloud data further includes: obtaining a point cloud map corresponding to the original point cloud data; matching the processed point cloud data with the point cloud map according to a preset matching algorithm to obtain the matching result of each processed point cloud data; and calculating the proportion of processed point cloud data whose matching result is a failed match to the total processed point cloud data to obtain a failed match percentage.

[0008] Optionally, in the fourth implementation of the first aspect of the present invention, the distortion compensation processing of the vehicle-removed point cloud data to obtain compensated point cloud data includes: dividing the vehicle-removed point cloud data into several continuous vehicle-removed point cloud data segments based on a preset time interval, wherein each vehicle-removed point cloud data segment has a corresponding intermediate time; traversing the vehicle-removed point cloud data segments frame by frame to determine the relative pose of any point cloud in each frame of the vehicle-removed point cloud data segment corresponding to the intermediate time; based on the relative pose, transforming the coordinates of any point cloud in the vehicle-removed point cloud data to the coordinates corresponding to the intermediate time, and summarizing to obtain compensated point cloud data.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, the step of performing rule detection and classification on the compensated point cloud data to obtain the point cloud category of each compensated point cloud data, and labeling the corresponding compensated point cloud data with the point cloud category to obtain processed point cloud data, includes: dividing the compensated point cloud data into at least two categories of classified point clouds according to corresponding rules based on the rule detection method; matching corresponding preset weights to the classified point clouds of different categories; and performing downsampling processing on the classified point clouds based on the preset weights to obtain processed point cloud data.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, determining the validity of the original point cloud data based on the ratio information and the deviation value includes: when the vehicle excluding ratio, the ground ratio, and the deviation value are all greater than a preset first threshold, determining that the collected data of the original point cloud data is normal; when the deviation value and the matching failure ratio are both less than a preset second threshold, determining that the collection environment of the original point cloud data is normal; and when both the collected data and the collection environment are normal, identifying the original point cloud data as valid.

[0011] A second aspect of the present invention provides a point cloud validity detection device, the point cloud validity detection device comprising: a point cloud preprocessing module for acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data; a ratio information calculation module for calculating the ratio information of various types of point cloud data in the processed point cloud data; a pose calculation module for identifying the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculating the deviation value between the initial pose and the optimized pose; and a validity judgment module for judging the validity of the raw point cloud data based on the ratio information and the deviation value.

[0012] Optionally, in the first implementation of the second aspect of the present invention, the point cloud preprocessing module is specifically used for: a point cloud vehicle removal unit, which performs vehicle removal processing on the original point cloud data to obtain vehicle-removed point cloud data; a point cloud compensation unit, which performs distortion compensation processing on the vehicle-removed point cloud data to obtain compensated point cloud data; and a point cloud classification unit, which performs rule detection and classification on the compensated point cloud data to obtain the point cloud category of each compensated point cloud data, and labels the corresponding compensated point cloud data with the point cloud category to obtain processed point cloud data.

[0013] Optionally, in a second implementation of the second aspect of the present invention, the proportion information calculation module is specifically used to: calculate the proportion of the vehicle-excluding point cloud data to the original point cloud data, to obtain a vehicle-excluding proportion value; and calculate the proportion of the processed point cloud data whose point cloud category is ground point cloud to the vehicle-excluding point cloud data, to obtain a ground proportion value.

[0014] Optionally, in a third implementation of the second aspect of the present invention, the proportion information calculation module is further configured to: obtain the point cloud map corresponding to the original point cloud data; match the processed point cloud data with the point cloud map according to a preset matching algorithm to obtain the matching result of each processed point cloud data; calculate the proportion of the processed point cloud data that failed to match to the total processed point cloud data, and obtain the proportion of the failed match value.

[0015] Optionally, in the fourth implementation of the second aspect of the present invention, the point cloud compensation unit is specifically used to: divide the vehicle-removing point cloud data into several continuous vehicle-removing point cloud data segments based on a preset time interval, wherein each vehicle-removing point cloud data segment has a corresponding intermediate time; traverse the vehicle-removing point cloud data segments frame by frame to determine the relative pose of any point cloud in each frame of the vehicle-removing point cloud data segment corresponding to the intermediate time; based on the relative pose, transform the coordinates of any point cloud in the vehicle-removing point cloud data to the coordinates corresponding to the intermediate time, and summarize them to obtain the compensated point cloud data.

[0016] Optionally, in a fifth implementation of the second aspect of the present invention, the point cloud classification unit is further configured to: classify the compensated point cloud data into at least two categories of classification point clouds according to corresponding rules based on a rule detection method; match corresponding preset weights for different categories of classification point clouds; and perform downsampling processing on the classification point clouds based on the preset weights to obtain processed point cloud data.

[0017] Optionally, in the sixth implementation of the second aspect of the present invention, the validity judgment module is specifically used to: determine that the collected data of the original point cloud data is normal when the vehicle ratio value, the ground ratio value, and the deviation value are all greater than a preset first threshold; determine that the collection environment of the original point cloud data is normal when the deviation value and the matching failure ratio value are both less than a preset second threshold; and identify the original point cloud data as valid when both the collected data and the collection environment are normal.

[0018] A third aspect of the present invention provides a point cloud validity detection device, comprising: a memory and at least one processor, wherein the memory stores a request, and the memory and the at least one processor are interconnected via a line; the at least one processor invokes the request in the memory to cause the point cloud validity detection device to perform the steps of the point cloud validity detection method described above.

[0019] A fourth aspect of the present invention provides a computer-readable storage medium storing a request that, when executed on a computer, causes the computer to perform the steps of the point cloud validity detection method described above.

[0020] In the technical solution of this invention, raw point cloud data is acquired and preprocessed to obtain processed point cloud data; the proportion information of various types of point cloud data in the processed point cloud data is calculated; the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data are identified, and the deviation value between the initial pose and the optimized pose is calculated; the validity of the raw point cloud data is determined based on the proportion information and the deviation value. This technical solution classifies the acquired raw point cloud data, determines several proportion information and deviation values ​​after classification, and identifies the environmental data at the time of raw point cloud data acquisition, thereby quickly determining the validity of the raw point cloud data and facilitating subsequent further processing of the raw point cloud data. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the first embodiment of the point cloud validity detection method in this invention;

[0022] Figure 2 This is a schematic diagram of the second embodiment of the point cloud validity detection method in this invention;

[0023] Figure 3 This is a schematic diagram of the third embodiment of the point cloud validity detection method in this invention;

[0024] Figure 4 This is a schematic diagram of one embodiment of the point cloud validity detection device in this invention;

[0025] Figure 5 This is a schematic diagram of another embodiment of the point cloud validity detection device in this invention;

[0026] Figure 6 This is a schematic diagram of one embodiment of the point cloud validity detection device in this invention. Detailed Implementation

[0027] In the technical solution of this invention, raw point cloud data is acquired and preprocessed to obtain processed point cloud data; the proportion information of various types of point cloud data in the processed point cloud data is calculated; the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data are identified, and the deviation value between the initial pose and the optimized pose is calculated; the validity of the raw point cloud data is determined based on the proportion information and the deviation value. This technical solution classifies the acquired raw point cloud data, determines several proportion information and deviation values ​​after classification, and identifies the environmental data at the time of raw point cloud data acquisition, thereby quickly determining the validity of the raw point cloud data and facilitating subsequent further processing of the raw point cloud data.

[0028] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar elements and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the point cloud validity detection method in this invention includes:

[0030] 101. Obtain the raw point cloud data and preprocess it to obtain processed point cloud data;

[0031] In this embodiment, the raw point cloud data refers to point cloud data collected by sensors installed on the vehicle within a preset time range, wherein the sensors include at least one of lidar, coordinate measuring machine, laser scanner or point cloud scanning system.

[0032] In practical applications, the raw point cloud data obtained by sensors installed on vehicles to collect the surrounding environment is not collected instantaneously during vehicle operation, but rather with a certain time delay. Within this time delay, factors such as vehicle displacement, other vehicles obstructing the sensors, or the sensors not effectively capturing the surrounding environment data in the raw point cloud data collected, necessitate preprocessing of the raw point cloud data to remove interference and obtain processed point cloud data.

[0033] Specifically, the preprocessing of raw point cloud data includes at least vehicle removal, distortion compensation, and downsampling.

[0034] 102. Calculate and process the proportion information of various types of point cloud data in the point cloud data;

[0035] In this embodiment, there are two parts to the calculation of the proportion of point cloud data.

[0036] Specifically, the first calculation of the proportion is for the point cloud after removing vehicles, which requires calculating the proportion of the point cloud data after removing vehicles to the original point cloud data, and the proportion of the ground point cloud data in the point cloud data after removing vehicles to the point cloud data after removing vehicles; the second calculation of the proportion is to match the processed point cloud data with the pre-stored point cloud map, and after matching, calculate the proportion of the processed point cloud data that failed to match to the total processed point cloud data.

[0037] 103. Identify the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculate the deviation between the initial pose and the optimized pose;

[0038] In this embodiment, for the obtained raw point cloud data, the corresponding initial pose and the optimized pose calculated from the processed point cloud data can be obtained. The deviation between the initial pose and the optimized pose is calculated by using the NDT matching algorithm, which serves as one of the criteria for judging the validity of the obtained raw point cloud data.

[0039] 104. Determine the validity of the original point cloud data based on the scale information and deviation value.

[0040] In this embodiment, the validity of the acquired raw point cloud data is determined by combining the obtained proportional information with the deviation value.

[0041] Specifically, when the vehicle proportion is less than 0.3, the ground proportion is greater than 0.7, and the deviation is greater than 0.5, the original point cloud data collected this time is considered to be surrounded by interference sources, and the NDT matching jump result is large, so the original point cloud data collected this time cannot be trusted.

[0042] On the other hand, when the failure rate is greater than 0.7 and the deviation is greater than 0.4, it is considered that the surrounding environment changed greatly during the collection process, resulting in the original point cloud data being unable to match the point cloud map or the original point cloud data not being used in the point cloud map. Therefore, the original point cloud data collected this time cannot be trusted.

[0043] On one hand, the method proposed in this application is based on the NDT (Normal Distribution Transformation) matching algorithm. It uses the original point cloud data of a single frame detected on the vehicle at the current moment, as well as the high-precision point cloud map stored on the vehicle in advance. The main idea of ​​the algorithm is to divide the laser point cloud of the current frame into grids of fixed size, and assume that the point cloud data distribution of each grid conforms to a normal distribution. The point cloud data of the current frame is transformed through the pose transformation matrix, and the joint Gaussian probability density of the transformed point cloud data and the point cloud map on the vehicle is summed. This probability sum can be maximized by the optimization method, thereby solving the transformation matrix and achieving the effect of inter-frame matching. Point clouds that cannot be matched are regarded as failed matching point clouds, and the matching failure value is calculated after traversing the matching.

[0044] In this embodiment, raw point cloud data is acquired and preprocessed to obtain processed point cloud data; the proportion information of various types of point cloud data in the processed point cloud data is calculated; the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data are identified, and the deviation value between the initial pose and the optimized pose is calculated; the validity of the raw point cloud data is determined based on the proportion information and the deviation value. The technical solution of this application classifies the acquired raw point cloud data, determines several proportion information and deviation values ​​after classification, and identifies the environmental data at the time of raw point cloud data acquisition, thereby quickly determining the validity of the raw point cloud data and facilitating subsequent further processing of the raw point cloud data.

[0045] Please see Figure 2 The second embodiment of the point cloud validity detection method in this invention includes:

[0046] 201. Perform vehicle removal processing on the original point cloud data to obtain vehicle-removed point cloud data;

[0047] In this embodiment, the system receives position information, contour information, and speed information of other vehicles around the vehicle detected by the sensor, and estimates multiple location regions that need to be removed. Then, it removes the point cloud data in these regions from the original point cloud data.

[0048] Specifically, by identifying vehicles in the original point cloud data, and based on parameters such as the time interval and spatial displacement between consecutive frames in the original point cloud data, the position, contour, and speed information of surrounding vehicles are determined. Using this determined position, contour, and speed information, the point cloud data requiring vehicle removal is selected, and the original point cloud data is then used to summarize the identified surrounding vehicles.

[0049] 202. Divide the vehicle point cloud data into several continuous vehicle point cloud data segments based on a preset time interval;

[0050] In this embodiment, the coordinates of each laser point in each frame of vehicle point cloud data are compensated. The compensation amount is the relative pose of the current laser point at that time (the vehicle pose at that time) relative to the middle time of that frame. First, the relative pose of the i-th point in each frame of point cloud relative to the middle time is calculated. Then, the coordinates of each laser point are transformed to the coordinates of the middle time of each frame of point cloud to obtain the compensated point cloud data.

[0051] Specifically, the formula for calculating relative pose is:

[0052]

[0053] 203. Traverse the vehicle-excluding point cloud data segment frame by frame to determine the relative pose of any point cloud in the vehicle-excluding point cloud data segment at the intermediate time.

[0054] 204. Based on relative pose, traverse the coordinates of any point cloud in the vehicle point cloud data and transform them to the coordinates corresponding to the intermediate time. After summarizing, the compensated point cloud data is obtained.

[0055] 205. Based on the rule-based detection method, the compensated point cloud data is divided into at least two categories of point clouds according to the corresponding rules;

[0056] In this embodiment, the compensated point cloud data obtained after vehicle removal processing and distortion compensation processing is classified based on the identified objects using a rule detection method. The data can be divided into ground point clouds and non-ground point clouds. The non-ground point clouds include distant point clouds, point clouds at the height above the vehicle body, and point clouds at the horizontal height of the vehicle body.

[0057] Specifically, for different types of point clouds, different weights will be assigned based on their type. Then, the point clouds classified according to different types will be downsampled with the assigned weights. The purpose is to reduce the proportion of the point cloud at the vehicle's horizontal height in the overall data, and obtain the downsampled point cloud data.

[0058] 206. Match corresponding preset weights to different categories of classification point clouds;

[0059] 207. Downsample the classified point cloud based on preset weights to obtain processed point cloud data;

[0060] In this embodiment, in order to improve the efficiency of point cloud processing, it is necessary to reduce the number of point clouds and improve the quality of point clouds, which can be achieved by downsampling and denoising the point clouds.

[0061] Specifically, since the vehicle horizontal height point cloud is generally collected from information of other vehicles in the surrounding area, and after vehicle removal processing, the corresponding vehicle horizontal height point cloud has greatly lost its reference value, the downsampling process focuses on reducing the corresponding weight of the vehicle horizontal height point cloud to effectively prevent data distortion.

[0062] 208. Calculate and process the proportion information of various types of point cloud data in the point cloud data;

[0063] 209. Identify the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculate the deviation between the initial pose and the optimized pose;

[0064] 210. Determine the validity of the original point cloud data based on the scale information and deviation value.

[0065] This embodiment, based on the previous embodiment, describes in detail the process of dividing the vehicle-removed point cloud data into several continuous segments based on a preset time interval, wherein each segment has a corresponding intermediate time. The process involves traversing each segment frame by frame to determine the relative pose of any point cloud within that segment to the intermediate time; based on the relative pose, transforming the coordinates of any point cloud in the segment to the coordinates corresponding to the intermediate time; and summarizing these coordinates to obtain the compensated point cloud data. Compared to traditional methods, this embodiment clarifies that for a single frame of raw point cloud data, the start and end times of the data acquisition are used to determine the corresponding intermediate time. Based on this intermediate time, the relative pose of any point in the raw point cloud data is determined to the intermediate time, and distortion compensation is applied to this point to obtain the compensated point cloud data. This reduces errors in subsequent determinations of the validity of the original point cloud.

[0066] Please see Figure 3 The third embodiment of the point cloud validity detection method in this invention includes:

[0067] 301. Obtain the raw point cloud data and preprocess it to obtain processed point cloud data;

[0068] 302. Calculate the proportion of vehicle-excluding point cloud data to the original point cloud data to obtain the vehicle-excluding proportion value;

[0069] In this embodiment, the vehicle point cloud data is N.remove_car The original point cloud data is N origin Except for the vehicle ratio value of R remove_car .

[0070] The specific formula for calculating the percentage of vehicles divided by the vehicle ratio is as follows:

[0071] R remove_car =N remove_car / N origin

[0072] 303. Calculate the proportion of processed point cloud data classified as ground point cloud to the total point cloud data excluding vehicle point cloud data to obtain the ground proportion value;

[0073] In this embodiment, the ground point cloud is N. ground The ground area ratio is R ground .

[0074] The specific formula for calculating the ground area ratio is as follows:

[0075] R ground =N ground / N remove_car

[0076] 304. Obtain the point cloud map corresponding to the raw point cloud data;

[0077] In this embodiment, the point cloud validity detection device that performs the point cloud validity detection method needs to preload a point cloud map corresponding to the location of the original point cloud data.

[0078] Specifically, the location of the original point cloud data can be obtained through technical means such as GPS satellite positioning system or manual annotation of the location where the original point cloud data was generated. Based on the location of the original point cloud data, a point cloud map corresponding to the location of the collection is loaded.

[0079] 305. According to the preset matching algorithm, the processed point cloud data is matched with the point cloud map to obtain the matching results of each processed point cloud data.

[0080] In this embodiment, the point cloud data is matched by loading a point cloud map corresponding to the collection location of the original point cloud data. That is, by identifying the ground, objects, buildings, environment and other information in the processed point cloud data, and comparing it with the ground, objects, buildings and environment and other information in the point cloud map, a matching result for the processed point cloud data is obtained.

[0081] Specifically, the comparison process determines whether there were any environmental changes during the acquisition of the original point cloud data.

[0082] 306. Calculate the proportion of point cloud data that failed to match out of all point cloud data to obtain the proportion of failed matches.

[0083] In this embodiment, the point cloud data is matched, and the matched point cloud data will be divided into two categories: successfully matched and unmatched. The ratio of the unmatched point cloud data to the total number of processed point cloud data is calculated to obtain the unmatched percentage.

[0084] Specifically, the success of the acquisition of raw point cloud data can be determined based on the proportion of failed matches, or the accuracy of the localization of the raw point cloud data can be determined. When the proportion of failed matches is lower than a preset threshold, it is considered that the raw point cloud data acquired this time can correspond to the preset point cloud map.

[0085] 307. Identify the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculate the deviation between the initial pose and the optimized pose;

[0086] 308. When the vehicle ratio, ground ratio, and deviation values ​​are all greater than the preset first threshold, the original point cloud data collection is determined to be normal.

[0087] In this embodiment, when the vehicle proportion value is less than 0.3, the ground proportion value is greater than 0.7, and the deviation value is greater than 0.5, the original point cloud data collected this time is considered to be surrounded by interference sources, and the NDT matching jump result is large, so the original point cloud data collected this time cannot be trusted.

[0088] Specifically, the original point cloud data can only be considered invalid if all of the above conditions are met.

[0089] 309. When both the deviation value and the proportion of matching failures are less than the preset second threshold, it is determined that the acquisition environment of the original point cloud data is normal.

[0090] In this embodiment, when both the matching failure rate and the deviation rate are greater than 0.7 and greater than 0.4, it is considered that the surrounding environment of the original point cloud data collected this time has changed greatly during the collection process, resulting in the original point cloud data being unable to match the point cloud map or the original point cloud data being not running in the point cloud map, and therefore the original point cloud data collected this time cannot be trusted.

[0091] 310. When both the collected data and the collection environment are normal, the raw point cloud data will be identified as valid.

[0092] In this embodiment, if the vehicle proportion value is less than 0.3, the ground proportion value is greater than 0.7, and the deviation value is greater than 0.5, or if the matching failure proportion value is greater than 0.7 and the deviation value is greater than 0.4, then the corresponding original point cloud data is invalid.

[0093] This embodiment, based on the previous embodiment, describes in detail the process of acquiring the point cloud map corresponding to the original point cloud data; matching the processed point cloud data with the point cloud map according to a preset matching algorithm to obtain the matching result of each processed point cloud data; and calculating the proportion of processed point cloud data that failed to match out of all processed point cloud data to obtain the matching failure percentage. Compared with traditional methods, this embodiment clarifies the processing procedure for the acquired original point cloud data. By matching the original point cloud data with the preset point cloud map, it can quickly determine whether the currently collected original point cloud data is similar to the pre-loaded point cloud map, and further determine the validity of the collected original point cloud data based on the point cloud map matching.

[0094] The point cloud validity detection method in the embodiments of the present invention has been described above. The point cloud validity detection device in the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 4 One embodiment of the point cloud validity detection device in this invention includes:

[0095] The point cloud preprocessing module 401 is used to acquire raw point cloud data and preprocess the raw point cloud data to obtain processed point cloud data.

[0096] The proportional information calculation module 402 is used to calculate the proportional information of various types of point cloud data in the processed point cloud data;

[0097] The pose calculation module 403 is used to identify the initial pose of the original point cloud data and the optimized pose of the processed point cloud data, and to calculate the deviation value between the initial pose and the optimized pose.

[0098] The validity judgment module 404 is used to judge the validity of the original point cloud data based on the ratio information and the deviation value.

[0099] In this embodiment of the invention, the point cloud validity detection device operates the aforementioned point cloud validity detection method, including: acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data; calculating the proportion information of various types of point cloud data in the processed point cloud data; identifying the initial pose of the raw point cloud data and the optimized pose of the processed point cloud data, and calculating the deviation value between the initial pose and the optimized pose; and determining the validity of the raw point cloud data based on the proportion information and the deviation value. The technical solution of this application classifies the acquired raw point cloud data, determines several proportion information and deviation values ​​after classification, and identifies the environmental data at the time of raw point cloud data acquisition, thereby quickly determining the validity of the raw point cloud data and facilitating subsequent further processing of the raw point cloud data.

[0100] Please see Figure 5The second embodiment of the point cloud validity detection device in this invention includes:

[0101] The cloud preprocessing module 401 is used to acquire raw point cloud data and preprocess the raw point cloud data to obtain processed point cloud data.

[0102] The proportional information calculation module 402 is used to calculate the proportional information of various types of point cloud data in the processed point cloud data;

[0103] The pose calculation module 403 is used to identify the initial pose of the original point cloud data and the optimized pose of the processed point cloud data, and to calculate the deviation value between the initial pose and the optimized pose.

[0104] The validity judgment module 404 is used to judge the validity of the original point cloud data based on the ratio information and the deviation value.

[0105] In this embodiment, the point cloud preprocessing module 401 is specifically used for:

[0106] The point cloud vehicle removal unit 4011 performs vehicle removal processing on the original point cloud data to obtain vehicle-removed point cloud data; the point cloud compensation unit 4012 performs distortion compensation processing on the vehicle-removed point cloud data to obtain compensated point cloud data; the point cloud classification unit 4013 performs rule detection and classification on the compensated point cloud data to obtain the point cloud category of each compensated point cloud data, and labels the corresponding compensated point cloud data with the point cloud category to obtain processed point cloud data.

[0107] In this embodiment, the ratio information calculation module 402 is specifically used for:

[0108] Calculate the proportion of the vehicle-excluding point cloud data to the original point cloud data to obtain the vehicle-excluding proportion value; calculate the proportion of the processed point cloud data whose point cloud category is ground point cloud to the vehicle-excluding point cloud data to obtain the ground proportion value.

[0109] In this embodiment, the ratio information calculation module 402 is further used for:

[0110] Obtain the point cloud map corresponding to the original point cloud data; match the processed point cloud data with the point cloud map according to the preset matching algorithm to obtain the matching result of each processed point cloud data; calculate the proportion of the processed point cloud data that failed to match to the total processed point cloud data to obtain the matching failure ratio.

[0111] In this embodiment, the point cloud compensation unit 4012 is specifically used for:

[0112] The vehicle removal point cloud data is divided into several continuous vehicle removal point cloud data segments based on a preset time interval. Each vehicle removal point cloud data segment has a corresponding intermediate time. The vehicle removal point cloud data segments are traversed frame by frame to determine the relative pose of any point cloud in each frame of the vehicle removal point cloud data segment with respect to the intermediate time. Based on the relative pose, the coordinates of any point cloud in the vehicle removal point cloud data are transformed to the coordinates corresponding to the intermediate time, and the results are summarized to obtain the compensated point cloud data.

[0113] In this embodiment, the point cloud classification unit 4013 is further used for:

[0114] The compensation point cloud data is divided into at least two categories of point clouds according to the corresponding rules based on the rule detection method; corresponding preset weights are matched for different categories of point clouds; and the point clouds are downsampled based on the preset weights to obtain processed point cloud data.

[0115] In this embodiment, the validity determination module 404 is specifically used for:

[0116] When the vehicle ratio, ground ratio, and deviation value are all greater than a preset first threshold, the original point cloud data is determined to be collected normally; when the deviation value and the matching failure ratio are both less than a preset second threshold, the original point cloud data is determined to be collected normally; when both the collected data and the collected environment are normal, the original point cloud data is identified as valid.

[0117] Based on the previous embodiment, this embodiment describes in detail the specific functions of each module and the unit composition of some modules. Through the above modules, the specific functions of the original modules are refined, the operation of the point cloud validity detection device is improved, its operational reliability is enhanced, and the actual logic between each step is clarified, thereby improving the practicality of the device.

[0118] above Figure 4 and Figure 5 The point cloud validity detection device in this embodiment of the invention is described in detail from the perspective of modular functional entities. The point cloud validity detection device in this embodiment of the invention is described in detail below from the perspective of hardware processing.

[0119] Figure 6This is a schematic diagram of a point cloud validity detection device 600 provided in an embodiment of the present invention. The point cloud validity detection device 600 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) for storing application programs 633 or data 632. The memory 620 and storage media 630 can be temporary or persistent storage. The program stored in the storage media 630 may include one or more modules (not shown in the diagram), each module may include a series of request operations on the point cloud validity detection device 600. Furthermore, the processor 610 may be configured to communicate with the storage media 630 and execute the series of request operations in the storage media 630 on the point cloud validity detection device 600 to implement the steps of the above-described point cloud validity detection method.

[0120] The point cloud validity testing device 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input / output interfaces 660, and / or one or more operating systems 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 6 The illustrated point cloud validity detection device structure does not constitute a limitation on the point cloud validity detection device provided in this application. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0121] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores a request that, when the request is executed on a computer, causes the computer to perform the steps of the point cloud validity detection method.

[0122] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0123] If the integrated unit is implemented as a software functional unit 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 invention, in essence, 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. This computer software product is stored in a storage medium and includes several requests to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0124] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A point cloud validity detection method, characterized in that, The point cloud validity detection method includes: The raw point cloud data is acquired and preprocessed to obtain processed point cloud data. Calculate the proportion information of various types of point cloud data in the processed point cloud data, the proportion information including the proportion of vehicle data, the proportion of ground data, and the proportion of matching failure data; Identify the initial pose of the original point cloud data and the optimized pose of the processed point cloud data, and calculate the deviation between the initial pose and the optimized pose; The validity of the original point cloud data is determined based on the ratio information and the deviation value. The determination of the validity of the original point cloud data based on the ratio information and the deviation value includes: When the vehicle excluding ratio, the ground ratio, and the deviation value are all greater than a preset first threshold, it is determined that the collected data of the original point cloud data is normal. When both the deviation value and the proportion of matching failures are less than a preset second threshold, it is determined that the acquisition environment of the original point cloud data is normal. When both the collected data and the collected environment are normal, the raw point cloud data is identified as valid. The process of acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data includes: performing vehicle removal processing on the raw point cloud data to obtain vehicle-removed point cloud data; performing distortion compensation processing on the vehicle-removed point cloud data to obtain compensated point cloud data; performing rule detection and classification on the compensated point cloud data to obtain point cloud categories for each compensated point cloud data, and labeling the corresponding compensated point cloud data with the point cloud categories to obtain processed point cloud data.

2. The point cloud validity detection method according to claim 1, characterized in that, The calculation of the proportion information of various types of point cloud data in the processed point cloud data includes: Calculate the proportion of the vehicle-excluding point cloud data to the original point cloud data to obtain the vehicle-excluding proportion value; The proportion of processed point cloud data classified as ground point cloud to the total point cloud data excluding vehicles is calculated to obtain the ground proportion value.

3. The point cloud validity detection method according to claim 2, characterized in that, The calculation of the proportion information of various types of point cloud data in the processed point cloud data also includes: Obtain the point cloud map corresponding to the original point cloud data; According to a preset matching algorithm, the processed point cloud data is matched with the point cloud map to obtain the matching results of each processed point cloud data. The proportion of point cloud data that failed to match is calculated out of all point cloud data to obtain the proportion of failed matches.

4. The point cloud validity detection method according to claim 1, characterized in that, The distortion compensation processing of the vehicle point cloud data to obtain compensated point cloud data includes: The vehicle removal point cloud data is divided into several continuous vehicle removal point cloud data segments based on a preset time interval, wherein each vehicle removal point cloud data segment has a corresponding intermediate time. The vehicle-excluding point cloud data segment is traversed frame by frame to determine the relative pose of any point cloud in the vehicle-excluding point cloud data segment and the intermediate time. Based on the relative pose, the coordinates of any point cloud in the vehicle-excluding point cloud data are transformed to the coordinates corresponding to the intermediate time, and the resulting data are aggregated to obtain the compensated point cloud data.

5. The point cloud validity detection method according to claim 2, characterized in that, The process of performing rule detection and classification on the compensated point cloud data to obtain the point cloud category of each compensated point cloud data, and labeling the corresponding compensated point cloud data with the point cloud category, to obtain processed point cloud data includes: Based on the rule-based detection method, the compensated point cloud data is divided into at least two categories of point clouds according to the corresponding rules. Match corresponding preset weights to different categories of point clouds; The classified point cloud is downsampled based on preset weights to obtain processed point cloud data.

6. A point cloud validity detection device, characterized in that, The point cloud validity detection device includes: The point cloud preprocessing module is used to acquire raw point cloud data and preprocess the raw point cloud data to obtain processed point cloud data. The proportion information calculation module is used to calculate the proportion information of various types of point cloud data in the processed point cloud data. The proportion information includes the proportion of vehicles, the proportion of ground, and the proportion of matching failures. The pose calculation module is used to identify the initial pose of the original point cloud data and the optimized pose of the processed point cloud data, and to calculate the deviation between the initial pose and the optimized pose. The validity determination module is used to determine the validity of the original point cloud data based on the ratio information and the deviation value; The determination of the validity of the original point cloud data based on the ratio information and the deviation value includes: When the vehicle excluding ratio, the ground ratio, and the deviation value are all greater than a preset first threshold, it is determined that the collected data of the original point cloud data is normal. When both the deviation value and the proportion of matching failures are less than a preset second threshold, it is determined that the acquisition environment of the original point cloud data is normal. When both the collected data and the collected environment are normal, the raw point cloud data is identified as valid. The process of acquiring raw point cloud data and preprocessing the raw point cloud data to obtain processed point cloud data includes: performing vehicle removal processing on the raw point cloud data to obtain vehicle-removed point cloud data; performing distortion compensation processing on the vehicle-removed point cloud data to obtain compensated point cloud data; performing rule detection and classification on the compensated point cloud data to obtain point cloud categories for each compensated point cloud data, and labeling the corresponding compensated point cloud data with the point cloud categories to obtain processed point cloud data.

7. A point cloud validity detection device, characterized in that, The point cloud validity detection device includes: a memory and at least one processor, wherein the memory stores requests, and the memory and the at least one processor are interconnected via a line; The at least one processor invokes the request in the memory to cause the point cloud validity detection device to perform the steps of the point cloud validity detection method as described in any one of claims 1-5.

8. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is executed by the processor, it implements each step of the point cloud validity detection method as described in any one of claims 1-5.

Images