Method and device for dividing ground points in laser radar point cloud, equipment and medium

Abstract

The invention discloses a method, a device and equipment for dividing ground points in a laser radar point cloud, and a medium. The method comprises the following steps: uploading the point cloud acquired by a laser radar to a GPU (Graphic Processing Unit) through a CUDA (Compute Unified Device Architecture); rasterizing the point cloud through the GPU, and determining a grid where each point in the point cloud is located; according to the method, the ground points in the grids are divided according to the z-axis coordinates of the points in the grids, the CUDA can process the calculation requirements of the multiple grids at the same time in a multi-thread mode, the time complexity brought by grid calculation can be effectively reduced, and therefore the real-time performance of dividing the ground points in the laser radar point cloud is effectively improved.

Description

Technical Field

[0001] The present application relates to the technical field of laser radar point cloud processing, and in particular to a method, device, equipment and medium for dividing ground points in a laser radar point cloud. Background Art

[0002] LiDAR is a remote sensing technology that uses laser pulses to measure distance. It detects the position, shape and structure of a target by emitting laser light at the target and measuring the reflected light. Point cloud is a form of LiDAR data. Autonomous vehicles need to accurately identify the road and ground conditions. Ground points can be used to detect road boundaries, lane lines, traffic signs and road conditions. Therefore, it is very important to accurately divide the ground points in the laser point cloud.

[0003] In the related technology, the methods for dividing ground points include plane grid method, point cloud normal vector method, model fitting method, facet grid method, deep learning method and other density-based, statistical characteristics-based and other division methods. The plane grid method is easy to implement and can cope with complex scenes, but the time complexity and the amount of calculation are very large when calculating each grid in turn, resulting in poor real-time performance; the point cloud normal vector method relies on the support of domain points, but it needs to be combined with other methods to cope with complex scenes; the model fitting method can better handle the division of ground points, but there are errors when dealing with complex terrains, and the calculation complexity is high, which may affect real-time performance; the facet grid method can better cope with the situation where there is curvature on the ground, but it is easy to have problems in scenes with a lot of noise, and the algorithm has high time complexity and needs to be optimized in real-time performance; the deep learning method can automatically extract complex features in the point cloud and accurately classify them, but it requires a large amount of training data and computing resources, and the interpretability of the model is relatively poor.

[0004] Therefore, how to accurately and real-time divide the ground points in the laser point cloud is a technical problem that needs to be solved urgently. Summary of the invention

[0005] The present application provides a method, device, equipment and medium for dividing ground points in a laser radar point cloud, which can solve the technical problem that the point cloud division method in the prior art cannot accurately and in real time divide the ground points in the laser point cloud.

[0006] In a first aspect, an embodiment of the present application provides a method for dividing ground points in a laser radar point cloud, wherein the method comprises:

[0007] The point cloud collected by the LiDAR is uploaded to the graphics processor GPU through the unified computing device architecture CUDA;

[0008] rasterizing the point cloud by the GPU, and determining the grid where each point in the point cloud is located;

[0009] The ground points in each grid are divided according to the z-axis coordinates of each point in each grid.

[0010] In combination with the first aspect, in one implementation, rasterizing the point cloud by the GPU and determining the grid where each point in the point cloud is located includes:

[0011] Determining the range of the point cloud by the GPU according to the point with the largest x-axis coordinate, the point with the smallest x-axis coordinate, the point with the largest y-axis coordinate, and the point with the smallest y-axis coordinate in the point cloud;

[0012] Based on a preset grid size and grid accuracy, and according to a range of the point cloud, rasterizing the point cloud;

[0013] Each grid is marked, and each point in the point cloud is marked to a corresponding grid according to the x-axis coordinate and the y-axis coordinate of each point.

[0014] In one embodiment, determining the ground points in each grid according to the z-axis coordinates of each point in each grid includes:

[0015] Sort the z-axis coordinates of the points in each grid respectively, and take the point with the smallest z-axis coordinate in each grid as the reference ground point of the corresponding grid;

[0016] Determine the distance between each point in each grid and the corresponding reference ground point;

[0017] Mark the points in each grid whose distance to the corresponding reference ground point is less than a preset distance threshold as ground points;

[0018] Points in each grid whose distance from the corresponding reference ground point is greater than or equal to the distance threshold are marked as non-ground points.

[0019] In one embodiment, before uploading the point cloud collected by the laser radar to the graphics processor GPU through the unified computing device architecture CUDA, it also includes:

[0020] De-distorting the point cloud using a preset algorithm;

[0021] Wherein, the preset algorithm includes a timestamp interpolation algorithm or a motion model linear interpolation algorithm.

[0022] In one embodiment, before uploading the point cloud collected by the laser radar to the graphics processor GPU through the unified computing device architecture CUDA, it also includes:

[0023] Stitch point clouds collected by multiple lidars in a multi-lidar system.

[0024] In one embodiment, before stitching point clouds collected by multiple laser radars in a multi-laser radar system, the method further includes:

[0025] Calibrate internal parameters and external parameters of multiple laser radars in the multi-laser radar system;

[0026] The internal parameters include lens distortion and measurement accuracy, and the external parameters include the position and direction of the reference coordinate system.

[0027] In one embodiment, after uploading the point cloud collected by the laser radar to the graphics processing unit GPU through the unified computing device architecture CUDA, and before rasterizing the point cloud through the GPU, the method further includes:

[0028] Filtering and downsampling the point cloud according to a preset filtering algorithm;

[0029] The filtering algorithm includes through filtering, voxel filtering or outlier filtering.

[0030] In a second aspect, an embodiment of the present application provides a device for dividing ground points in a laser radar point cloud, wherein the device for dividing ground points in a laser radar point cloud comprises:

[0031] An upload module is used to upload the point cloud collected by the lidar to the graphics processor GPU through the unified computing device architecture CUDA;

[0032] A processing module, which is used to rasterize the point cloud through the GPU and determine the grid where each point in the point cloud is located;

[0033] The division module is used to divide the ground points in each grid according to the z-axis coordinates of each point in each grid.

[0034] In a third aspect, an embodiment of the present application provides a device for dividing ground points in a laser radar point cloud, wherein the device for dividing ground points in a laser radar point cloud comprises a processor, a memory, and a laser radar point cloud ground point program stored in the memory and executable by the processor, wherein when the laser radar point cloud ground point program is executed by the processor, the steps of the method for dividing ground points in a laser radar point cloud as described in any one of the above items are implemented.

[0035] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a ground point program in a laser radar point cloud is stored, wherein when the ground point program in the laser radar point cloud is executed by a processor, the steps of the method for dividing ground points in a laser radar point cloud as described in any of the above items are implemented.

[0036] The embodiments of the present application provide a method, apparatus, device and medium for dividing ground points in a laser radar point cloud. The point cloud collected by the laser radar is uploaded to a graphics processing unit (GPU) through a unified computing device architecture (CUDA). The point cloud is rasterized by the GPU, and the grid where each point in the point cloud is located is determined. The ground points in each grid are divided according to the z-axis coordinates of each point in each grid. CUDA can process the computing requirements of multiple grids simultaneously through multi-threading, which can effectively reduce the time complexity brought by grid computing, thereby effectively improving the real-time performance of dividing the ground points in the laser radar point cloud. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A schematic diagram of the process of dividing ground points in the laser radar point cloud of this application;

[0038] Figure 2 A schematic diagram of a specific process of dividing ground points in a laser radar point cloud in this application;

[0039] Figure 3 This is a schematic diagram of the functional modules of the device for dividing ground points in the laser radar point cloud of this application;

[0040] Figure 4 This is a schematic diagram of the hardware structure of a device for dividing ground points in a laser radar point cloud involved in an embodiment of the present application. DETAILED DESCRIPTION

[0041] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0042] First, some technical terms in the present application are explained to facilitate those skilled in the art to understand the present application.

[0043] CUDA (Compute Unified Device Architecture) is a parallel computing platform and programming model that can achieve parallel processing through multi-threading.

[0044] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0045] In a first aspect, an embodiment of the present application provides a method for dividing ground points in a laser radar point cloud.

[0046] In one embodiment, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the method for dividing ground points in a laser radar point cloud of the present application. Figure 1 As shown in FIG. 1 , the method for dividing ground points in the lidar point cloud includes:

[0047] Step S101: upload the point cloud collected by the laser radar to the graphics processing unit GPU through the unified computing device architecture CUDA.

[0048] It is worth mentioning that after the lidar collects the point cloud, it will first upload the point cloud to the CPU, and then upload the point cloud from the CPU to the GPU through CUDA. In the subsequent steps, the point cloud will be accelerated in the GPU through multi-threading.

[0049] As a preferred embodiment, Figure 2 As shown, before uploading the point cloud to the GPU, it also includes: de-distorting the point cloud by a preset algorithm. The physical characteristics, environmental factors and motion dynamics of the LiDAR sensor itself will cause point cloud distortion. According to the LiDAR model and imaging principle, the point cloud can be de-distorted by using de-distortion algorithms such as timestamp interpolation or linear interpolation of motion models, thereby eliminating or reducing errors caused by LiDAR sensor hardware or external environmental factors. A vehicle's radar system may include multiple types of LiDARs. Therefore, different algorithms can be used to de-distort point clouds collected by different types of LiDARs in a multi-LiDAR system.

[0050] In one embodiment, before uploading the point cloud collected by the laser radar to the graphics processor GPU, the step also includes: stitching the point clouds collected by multiple laser radars in the multi-laser radar system.

[0051] Preferably, before stitching the point clouds collected by multiple laser radars in the multi-laser radar system, the method further includes: calibrating the internal and external parameters of the multiple laser radars in the multi-laser radar system; wherein the internal parameters include parameters such as lens distortion and measurement accuracy, and the external parameters include parameters such as the position and direction of the reference coordinate system. Calibration between multiple laser radars can ensure that the ground point part of the point cloud collected by the laser radars will not be delaminated after stitching.

[0052] Step S102: rasterize the point cloud by using the GPU, and determine the grid where each point in the point cloud is located.

[0053] Specifically, the GPU determines the range of the point cloud according to the point with the largest x-axis coordinate, the point with the smallest x-axis coordinate, the point with the largest y-axis coordinate, and the point with the smallest y-axis coordinate in the point cloud; based on a preset grid size and grid accuracy, the point cloud is rasterized according to the range of the point cloud; each grid is marked, and each point in the point cloud is marked to a corresponding grid according to the x-axis coordinate and y-axis coordinate of each point.

[0054] It is worth noting that the grid size is the physical size of the grid, which can be determined according to the scale of the point cloud. The grid accuracy includes the grid resolution, which can determine the level of detail of the rasterized result. The grid size and accuracy are strongly related to the operating efficiency, and can be weighed and adjusted according to application requirements and computing resources.

[0055] In an optional embodiment, after uploading the point cloud collected by the lidar to the graphics processing unit GPU through the unified computing device architecture CUDA, before rasterizing the point cloud through the GPU, it also includes: filtering and downsampling the point cloud according to a preset filtering algorithm; wherein the filtering algorithm includes pass-through filtering, voxel filtering or outlier filtering. By filtering and downsampling the point cloud, noise points in the point cloud can be effectively removed, the algorithm operation stability can be improved, the point cloud density can be reduced, and the computational overhead can be reduced. The discrete points to be removed during the downsampling process are especially some obvious noise points below the ground, otherwise some ground points at the noise location may be mistakenly marked as non-ground points.

[0056] Step S103: Divide the ground points in each grid according to the z-axis coordinates of each point in each grid.

[0057] Specifically, the z-axis coordinates of the points in each grid are sorted respectively, and the point with the smallest z-axis coordinate in each grid is used as the reference ground point of the corresponding grid; the distance between each point in each grid and the corresponding reference ground point is determined; the points in each grid whose distance to the corresponding reference ground point is less than a preset distance threshold are marked as ground points; and the points in each grid whose distance to the corresponding reference ground point is greater than or equal to the distance threshold are marked as non-ground points.

[0058] It is worth noting that the distance threshold can be set according to the needs. The Z-axis coordinate of each point in the point cloud represents the coordinate of the point in the vertical direction, that is, the height of the point. The point with the smallest z-axis coordinate in a grid is the point with the lowest height. This point can be considered as a ground point and used as the reference ground point. Points whose distance to the reference ground point is less than the distance threshold indicate that their height is also very low, so they can be recorded as ground points. Points whose distance to the reference ground point is greater than or equal to the distance threshold indicate that their height is relatively high, so they can be recorded as non-ground points, thereby completing the division of ground points and non-ground points in the point cloud.

[0059] Furthermore, after the GPU completes the marking of ground points and non-ground points in the point cloud, the marked point cloud is retrieved from the GPU to the CPU using CUDA and passed downstream.

[0060] Explanatory, rasterizing the point cloud to divide the ground points and non-ground points in the point cloud can effectively improve the scene adaptability and robustness of the lidar point cloud ground segmentation. Compared with the method based on model fitting, it is better for complex and uneven roads. Since the grid method involves a large number of grid data points, each grid needs to be calculated multiple times, which will cause a large amount of calculation to affect the real-time performance of ground point division. CUDA can start multiple threads in parallel to process the calculation of each point in each grid at the same time, thereby greatly reducing the total calculation time. The CUDA acceleration solution is very consistent with the algorithm logic of the grid method. Its multi-threaded acceleration solution can effectively reduce the time complexity brought by the grid method and effectively improve the real-time performance of this case.

[0061] In a second aspect, an embodiment of the present application also provides a device for dividing ground points in a laser radar point cloud.

[0062] In one embodiment, referring to Figure 3 , Figure 3 This is a functional module diagram of an embodiment of a device for dividing ground points in a laser radar point cloud of the present application. Figure 3 As shown, the device for dividing ground points in the laser radar point cloud includes:

[0063] An upload module is used to upload the point cloud collected by the lidar to the graphics processor GPU through the unified computing device architecture CUDA;

[0064] A processing module, which is used to rasterize the point cloud through the GPU and determine the grid where each point in the point cloud is located;

[0065] The division module is used to divide the ground points in each grid according to the z-axis coordinates of each point in each grid.

[0066] Furthermore, in one embodiment, the processing module is further configured to:

[0067] Determining the range of the point cloud by the GPU according to the point with the largest x-axis coordinate, the point with the smallest x-axis coordinate, the point with the largest y-axis coordinate, and the point with the smallest y-axis coordinate in the point cloud;

[0068] Based on a preset grid size and grid accuracy, and according to a range of the point cloud, rasterizing the point cloud;

[0069] Each grid is marked, and each point in the point cloud is marked to a corresponding grid according to the x-axis coordinate and the y-axis coordinate of each point.

[0070] Furthermore, in one embodiment, the partitioning module is also used for:

[0071] Sort the z-axis coordinates of the points in each grid respectively, and take the point with the smallest z-axis coordinate in each grid as the reference ground point of the corresponding grid;

[0072] Determine the distance between each point in each grid and the corresponding reference ground point;

[0073] Mark the points in each grid whose distance to the corresponding reference ground point is less than a preset distance threshold as ground points;

[0074] Points in each grid whose distance from the corresponding reference ground point is greater than or equal to the distance threshold are marked as non-ground points.

[0075] Furthermore, in one embodiment, the device is also used for:

[0076] Before uploading the point cloud collected by the laser radar to the graphics processing unit GPU through the unified computing device architecture CUDA, the point cloud is dedistorted by a preset algorithm;

[0077] Wherein, the preset algorithm includes a timestamp interpolation algorithm or a motion model linear interpolation algorithm.

[0078] Furthermore, in one embodiment, the device is also used for:

[0079] Before uploading the point cloud collected by the lidar to the graphics processing unit GPU through the unified computing device architecture CUDA, the point clouds collected by multiple lidars in the multi-lidar system are spliced.

[0080] Furthermore, in one embodiment, the device is also used for:

[0081] Before stitching point clouds collected by multiple laser radars in the multi-laser radar system, calibrate internal and external parameters of multiple laser radars in the multi-laser radar system;

[0082] The internal parameters include lens distortion and measurement accuracy, and the external parameters include the position and direction of the reference coordinate system.

[0083] Furthermore, in one embodiment, the device is also used for:

[0084] After uploading the point cloud collected by the lidar to the graphics processing unit GPU through the unified computing device architecture CUDA, before rasterizing the point cloud through the GPU, the point cloud is filtered and downsampled according to a preset filtering algorithm; wherein the filtering algorithm includes pass-through filtering, voxel filtering or outlier filtering.

[0085] Among them, the functional implementation of each module in the above-mentioned device for dividing ground points in the laser radar point cloud corresponds to the various steps in the above-mentioned embodiment of the method for dividing ground points in the laser radar point cloud, and its functions and implementation processes will not be repeated here one by one.

[0086] In a third aspect, an embodiment of the present application provides a device for dividing ground points in a laser radar point cloud. The device for dividing ground points in a laser radar point cloud can be a personal computer (PC), a laptop computer, a server, or other device with data processing capabilities.

[0087] Reference Figure 4 , Figure 4 Schematic diagram of the hardware structure of the device for dividing ground points in the laser radar point cloud involved in the embodiment of the present application. In the embodiment of the present application, the device for dividing ground points in the laser radar point cloud may include a processor, a memory, a communication interface and a communication bus.

[0088] The communication bus may be of any type and is used to interconnect the processor, the memory, and the communication interface.

[0089] The communication interface includes input / output (I / O) interface, physical interface and logical interface, etc., which are used to realize the interconnection of devices inside the device for dividing ground points in the laser radar point cloud, and the interface for realizing the interconnection between the device for dividing ground points in the laser radar point cloud and other devices (such as other computing devices or user devices). The physical interface can be an Ethernet interface, a fiber optic interface, an ATM interface, etc.; the user device can be a display, a keyboard, etc.

[0090] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0091] The processor may be a general-purpose processor, which may call the ground point division program in the laser radar point cloud stored in the memory and execute the ground point division method in the laser radar point cloud provided in the embodiment of the present application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the ground point division program in the laser radar point cloud is called may refer to the various embodiments of the ground point division method in the laser radar point cloud of the present application, which will not be described in detail here.

[0092] Those skilled in the art will understand that Figure 4 The hardware structure shown in the figure does not constitute a limitation on the present application, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0093] In a fourth aspect, an embodiment of the present application also provides a computer-readable storage medium.

[0094] The computer-readable storage medium of the present application stores a program for dividing ground points in a laser radar point cloud, wherein when the program for dividing ground points in a laser radar point cloud is executed by a processor, the steps of the method for dividing ground points in a laser radar point cloud as described above are implemented.

[0095] Among them, the method implemented when the ground point division program in the laser radar point cloud is executed can refer to the various embodiments of the ground point division method in the laser radar point cloud of the present application, and will not be repeated here.

[0096] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are only for description and do not represent the advantages or disadvantages of the embodiments.

[0097] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit "first", "second" and "third" to different types.

[0098] In the description of the embodiments of the present application, "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a specific way.

[0099] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; the “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0100] In some processes described in the embodiments of the present application, multiple operations or steps that appear in a specific order are included, but it should be understood that these operations or steps may not be executed in the order in which they appear in the embodiments of the present application or in parallel, and the sequence number of the operation is only used to distinguish the different operations, and the sequence number itself does not represent any execution order. In addition, these processes may include more or fewer operations, and these operations or steps may be executed in sequence or in parallel, and these operations or steps may be combined.

[0101] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD) as described above, and includes a number of instructions for a terminal device to execute the methods described in each embodiment of the present application.

[0102] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for dividing ground points in a laser radar point cloud, characterized in that: The method for dividing ground points in the laser radar point cloud includes: The point cloud collected by the LiDAR is uploaded to the graphics processor GPU through the unified computing device architecture CUDA; rasterizing the point cloud by the GPU, and determining the grid where each point in the point cloud is located; The ground points in each grid are divided according to the z-axis coordinates of each point in each grid.

2. The method for dividing ground points in a laser radar point cloud according to claim 1, characterized in that: Rasterizing the point cloud by the GPU and determining the grid where each point in the point cloud is located, including: Determining the range of the point cloud by the GPU according to the point with the largest x-axis coordinate, the point with the smallest x-axis coordinate, the point with the largest y-axis coordinate, and the point with the smallest y-axis coordinate in the point cloud; Based on a preset grid size and grid accuracy, and according to a range of the point cloud, rasterizing the point cloud; Each grid is marked, and each point in the point cloud is marked to a corresponding grid according to the x-axis coordinate and the y-axis coordinate of each point.

3. The method for dividing ground points in a laser radar point cloud according to claim 2, characterized in that: According to the z-axis coordinates of each point in each grid, the ground points in each grid are determined, including: Sort the z-axis coordinates of the points in each grid respectively, and take the point with the smallest z-axis coordinate in each grid as the reference ground point of the corresponding grid; Determine each point in each grid and the corresponding reference ground point; Mark the points in each grid whose distance to the corresponding reference ground point is less than a preset distance threshold as ground points; Points in each grid whose distance from the corresponding reference ground point is greater than or equal to the distance threshold are marked as non-ground points.

4. The method for dividing ground points in a laser radar point cloud according to claim 1, characterized in that: Before uploading the point cloud collected by the lidar to the graphics processor GPU through the unified computing device architecture CUDA, it also includes: De-distorting the point cloud using a preset algorithm; Wherein, the preset algorithm includes a timestamp interpolation algorithm or a motion model linear interpolation algorithm.

5. The method for dividing ground points in a laser radar point cloud according to claim 1, characterized in that: Before uploading the point cloud collected by the lidar to the graphics processor GPU through the unified computing device architecture CUDA, it also includes: Stitch point clouds collected by multiple lidars in a multi-lidar system.

6. The method for dividing ground points in a laser radar point cloud according to claim 5, characterized in that: Before stitching the point clouds collected by multiple lidars in a multi-lidar system, it also includes: Calibrate internal parameters and external parameters of multiple laser radars in the multi-laser radar system; The internal parameters include lens distortion and measurement accuracy, and the external parameters include the position and direction of the reference coordinate system.

7. The method for dividing ground points in a laser radar point cloud according to claim 1, after uploading the point cloud collected by the laser radar to a graphics processing unit (GPU) through a unified computing device architecture (CUDA), and before rasterizing the point cloud through the GPU, further comprising: Filtering and downsampling the point cloud according to a preset filtering algorithm; The filtering algorithm includes through filtering, voxel filtering or outlier filtering.

8. A device for dividing ground points in a laser radar point cloud, characterized in that: The device for dividing ground points in the laser radar point cloud comprises: An upload module is used to upload the point cloud collected by the lidar to the graphics processor GPU through the unified computing device architecture CUDA; A processing module, which is used to rasterize the point cloud through the GPU and determine the grid where each point in the point cloud is located; The division module is used to divide the ground points in each grid according to the z-axis coordinates of each point in each grid.

9. A device for dividing ground points in a laser radar point cloud, characterized in that: The device for dividing ground points in a laser radar point cloud comprises a processor, a memory, and a program for dividing ground points in a laser radar point cloud stored in the memory and executable by the processor, wherein when the program for dividing ground points in a laser radar point cloud is executed by the processor, the steps of the method for dividing ground points in a laser radar point cloud as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a ground point program in a laser radar point cloud, wherein when the ground point program in a laser radar point cloud is executed by a processor, the steps of the method for dividing ground points in a laser radar point cloud as described in any one of claims 1 to 7 are implemented.

Images