Ground point cloud segmentation method, device, equipment and medium

Abstract

Embodiments of the present disclosure disclose a ground point cloud segmentation method, device, equipment and medium, the method comprising: determining a plane ROI according to the distribution characteristics of a current frame point cloud and / or corresponding road network information; dividing a plurality of containers according to different height intervals for the same plane ROI; respectively counting point cloud points in the current frame point cloud falling in each container; determining interval scores of each container according to the point cloud points in each container, wherein the higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points; and determining ground points according to the container with the highest interval score. The method does not require a large amount of labeled data, has a low computational complexity, can be implemented on any low-power platform, and can be adapted to various mechanical or solid-state laser radars with different beams.

Description

Technical Field

[0001] This disclosure relates to the field of point cloud segmentation technology, and in particular to a method, apparatus, device and medium for ground point cloud segmentation. Background Technology

[0002] Ground point cloud segmentation is an important task in point cloud processing, which aims to separate ground points from non-ground points. It is commonly used in fields such as autonomous driving, building modeling, and terrain reconstruction.

[0003] Currently, commonly used ground point cloud segmentation methods include the following: 1. RANSAC-based plane fitting method: This is a commonly used ground point cloud segmentation method. The idea is to randomly sample a point set based on the RANSAC algorithm and estimate the ground plane model through plane fitting. This method is simple and easy to implement, but its segmentation effect is poor for non-flat terrain. 2. Machine learning-based classification method: This method uses machine learning algorithms, such as support vector machines and neural networks, to classify point clouds into ground points and non-ground points. This method requires a large amount of training data, but the classification effect is good. 3. Elevation threshold method: This method is simple and intuitive. Ground elevation is usually used as a threshold; points above the threshold are considered non-ground points, and points below the threshold are considered ground points. However, this method may misclassify points in terrain with large undulations. 4. Clustering-based method: This method finds dense regions in the point cloud and considers these regions as ground points. Commonly used clustering algorithms include K-means and DBSCAN. This method is suitable for irregular or complex terrain, but its computational complexity is high and it is limited by the sparsity of different LiDAR systems. 5. Laser Waveform-Based Method: This method uses the waveform information emitted by the laser to classify ground and non-ground points. This method can effectively avoid situations where ground texture is unclear, but it requires high-precision modeling of the LiDAR beam and is not suitable for solid-state LiDAR.

[0004] In view of this, the present invention is hereby proposed. Summary of the Invention

[0005] To solve the above-mentioned technical problems, or at least partially solve them, this disclosure provides a ground point cloud segmentation method, apparatus, device, and medium. This method does not require the labeling of a large amount of data and has low computational complexity. It can be implemented on any low-computing-power platform and is adaptable to various mechanical or solid-state lidars with different wire beams.

[0006] In a first aspect, embodiments of this disclosure provide a ground point cloud segmentation method, the method comprising:

[0007] Determine the planar ROI based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information;

[0008] Divide the same planar ROI into multiple containers according to different height ranges;

[0009] Count the point cloud points in the current frame that fall into each of the aforementioned containers;

[0010] The interval score of each container is determined based on the point cloud points in each container. The higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points.

[0011] The ground point is determined based on the container with the highest interval score.

[0012] Secondly, embodiments of this disclosure also provide a ground point cloud segmentation device, the device comprising:

[0013] The first determining module is used to determine the planar ROI based on the distribution characteristics of the point cloud in the current frame and / or the corresponding road network information;

[0014] The partitioning module is used to divide the same planar ROI into multiple containers according to different height ranges;

[0015] The statistics module is used to count the point cloud points that fall into each of the containers in the current frame point cloud;

[0016] The second determining module is used to determine the interval score of each container based on the point cloud points in each container. The higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points.

[0017] The third determination module is used to determine the ground point based on the container with the highest interval score.

[0018] Thirdly, this disclosure also provides an electronic device, which includes: one or more processors; a storage device for storing one or more programs; and when the one or more programs are executed by the one or more processors, the one or more processors implement the ground point cloud segmentation method as described above.

[0019] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the ground point cloud segmentation method as described above.

[0020] This disclosure provides a ground point cloud segmentation method. First, a planar Region of Interest (ROI) is determined based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information. Then, the same planar ROI is divided into multiple containers according to different height intervals. Next, the point cloud points falling into each container in the current frame are counted. An interval score is determined for each container based on the point cloud points in each container; the higher the interval score, the greater the probability that the point cloud point in the corresponding container is a ground point. The ground point is determined based on the container with the highest interval score. This method achieves ground point cloud segmentation and requires no large amount of labeled data, has low computational complexity, can be implemented on any low-computing-power platform, and is adaptable to various types of mechanical or solid-state lidar with different beamwidths. Attached Figure Description

[0021] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0022] Figure 1 This is a flowchart of a ground point cloud segmentation method according to an embodiment of the present disclosure;

[0023] Figure 2 This is a schematic diagram of a planar region of interest (ROI) according to an embodiment of this disclosure;

[0024] Figure 3 This is a schematic diagram of the structure of a ground point cloud segmentation device according to an embodiment of the present disclosure;

[0025] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. Detailed Implementation

[0026] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0027] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0028] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0029] Figure 1 This is a flowchart illustrating a ground point cloud segmentation method according to an embodiment of this disclosure. The method can be executed by a ground point cloud segmentation device, which can be implemented in software and / or hardware and can be configured in an electronic device. Figure 1 As shown, the method may specifically include the following steps:

[0030] S110. Determine the planar ROI based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information.

[0031] Specifically, if the point cloud in the current frame is relatively sparse, the determined planar ROI (Region of Interest) should be as large as possible so that the planar ROI includes as many point clouds as possible; if the point cloud in the current frame is relatively dense, the planar ROI can be smaller.

[0032] For example, see references to Figure 2 The diagram shows a planar ROI, which includes two planar ROIs: a first planar ROI210 and a second planar ROI220.

[0033] Furthermore, the boundaries of a planar Region of Interest (ROI) can be determined based on known road network information. Specifically, more accurate boundary information is obtained from the road network information, and then polygonal ROIs are defined. For example, if six boundary points p1, p2, p3, p4, p5, and p6 are obtained from the road network information, then the hexagonal region formed by connecting these six boundary points is a planar ROI. As another example, if six boundary points p1, p2, p3, p4, and p5 are obtained from the road network information, then the pentagonal region formed by connecting these five boundary points is a planar ROI.

[0034] S120. Divide the same planar ROI into multiple containers according to different height ranges.

[0035] In three-dimensional space, the plane containing the ROI is labeled xoy. Along the z-direction, the same ROI is divided into multiple containers according to different height intervals. For example, the height interval (0, 1) corresponds to one container, and the height interval (1, 2) corresponds to another container.

[0036] S130. Count the point cloud points that fall into each of the containers in the current frame point cloud.

[0037] Specifically, the point cloud points falling into each container are counted based on the coordinates of each point cloud point.

[0038] S140. Determine the interval score of each container based on the point cloud points in each container. The higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points.

[0039] For example, determining the interval score of each container based on the point cloud points in each container includes:

[0040] For containers in different height ranges on the same plane ROI, starting from the container with the lowest height range, the range score of each container is determined based on the total number of point cloud points in the container and the preset reward rules.

[0041] For example, determining the interval score for each container based on the total number of point cloud points in the container and a preset reward rule includes:

[0042] The reward points for the current container are determined based on the preset reward rules;

[0043] The sum of the total number of point cloud points in the current container and the reward points is determined as the interval score of the current container.

[0044] The step of determining the reward points for the current container based on the preset reward rule includes:

[0045] If the total number of point cloud points in the current container is greater than the interval score of the previous container adjacent to the current container, and / or if the range of continuous distribution of point cloud points in the current container reaches a first threshold, then the current container is determined to receive reward points; wherein, the number of reward points is determined based on the total number of point cloud points in the current container, and / or based on the height interval corresponding to the current container, the more point cloud points in the current container, the more reward points; the higher the height interval corresponding to the current container, the fewer reward points.

[0046] Suppose a planar Region of Interest (ROI) is divided into four containers based on different height intervals. The first container corresponds to a height interval of (0, 1), the second to (1, 2), the third to (2, 3), and the fourth to (3, 4). Starting with the container with the lowest height interval, the interval score for each container is determined based on the total number of point cloud points in the container and a preset reward rule. Specifically, the reward points for the first container are determined as follows: if the total number of point cloud points in the first container is greater than the interval score of the container adjacent to it, then the first container receives reward points. Since there is no container adjacent to the first container, or its interval score is zero, the first container receives reward points. The sum of the reward points and the total number of point cloud points in the first container is the interval score of the first container. This is how the interval score of the first container is determined. Next, the second container is taken as the current container, and the interval score of the second container is determined. The previous container adjacent to the second container is the first container. If the total number of point cloud points in the second container is greater than the interval score of the first container, the second container is determined to receive reward points. This process is repeated to determine the interval score of each container.

[0047] The number of reward points is determined based on the total number of point cloud points in the current container and / or the height range corresponding to the current container. The more point cloud points in the current container, the more reward points are awarded; conversely, the higher the height range corresponding to the current container, the fewer reward points are awarded. Since ground point clouds are relatively concentrated, the height range of a container with a larger total number of point cloud points is more likely to be a range of ground point clouds. Similarly, the higher the corresponding height range, the lower the probability of it being a range of ground point clouds; therefore, the higher the corresponding height range, the fewer reward points are awarded.

[0048] In general, determining the interval score of each container based on the point cloud points in each container includes:

[0049] For the current container in each container, the interval score of the current container is determined based on the total number of point cloud points in the current container, and / or the interval score of the containers adjacent to the current container, and / or the range of continuous distribution of point cloud points in the current container.

[0050] The range of continuous distribution of point cloud points within the current container can be characterized by the distances between the point cloud points. For example, a point can be selected as the center point from the point cloud points in the current container, and the distances between the center point and other points can be calculated. The number of points whose distances are less than a set threshold can be counted. The range of continuous distribution of point cloud points is represented by the number of such points; the more such points there are, the larger the range of continuous distribution of point cloud points. Optionally, the range of continuous distribution of point cloud points can also be represented by the average distance between two adjacent points or the sum of the distances between two adjacent points.

[0051] S150. Determine the ground point based on the container with the highest interval score.

[0052] In some implementations, the point cloud points in the container with the highest interval score can be directly identified as ground points.

[0053] In other implementations, to find more ground points and improve the segmentation accuracy of ground points, the height range of the container with the highest interval score can be appropriately expanded. For example, determining the ground point based on the container with the highest interval score includes:

[0054] Determine whether the similarity between the point cloud features in the target container and the point cloud features in the container with the highest interval score is less than a similarity threshold, wherein the target container is the container adjacent to the container with the highest interval score;

[0055] If the similarity is less than the similarity threshold and the number of target containers is one, determine whether the target container is above or below the container with the highest score in the interval;

[0056] If the target container is above the container with the highest interval score, the upper limit of the height interval corresponding to the container with the highest interval score is expanded. If the target container is below the container with the highest interval score, the lower limit of the height interval corresponding to the container with the highest interval score is expanded to obtain the height interval after the first expansion. If the similarity is less than the similarity threshold and there are two target containers, the upper and lower limits of the height interval corresponding to the container with the highest interval score are expanded simultaneously to obtain the height interval after the first expansion.

[0057] The lower and upper limits of the height range after the first expansion are expanded by a set value to obtain the height range after the second expansion.

[0058] Point cloud points falling within the second expanded altitude range are identified as ground points.

[0059] Assuming the height interval corresponding to the container with the highest interval score is (-0.1, 0), and the height interval corresponding to the target container is (0, 0.1), meaning the target container is above the container with the highest interval score, if the similarity is less than a similarity threshold, the upper limit of the height interval corresponding to the container with the highest interval score is expanded. The specific expansion amount can be determined empirically; for example, expanding by 0.1 results in a height interval of (-0.1, 0.1) after the first expansion. Considering the layering of the ground, the lower and upper limits of the height interval after the first expansion are expanded by set values. These set values ​​can be determined empirically or by engineers based on the ground features in the application scenario. For example, if the set value is 0.1, the height interval after the second expansion is (-0.2, 0.2), and points falling within (-0.2, 0.2) are identified as ground points. This improves the segmentation accuracy of ground points.

[0060] Assuming the height range corresponding to the container with the highest interval score is (-0.1, 0), and the height range corresponding to the target container is (-0.2, -0.1), meaning the target container is below the container with the highest interval score, if the similarity is less than the similarity threshold, then the lower limit of the height range corresponding to the container with the highest interval score is expanded, for example, by 0.1. Then the height range after the first expansion is (-0.2, 0).

[0061] Assuming the height interval corresponding to the container with the highest interval score is (-0.1, 0), and there are two target containers, one of which has a height interval of (0, 0.1), meaning it is above the container with the highest interval score, and the other has a height interval of (-0.2, -0.1), meaning it is below the container with the highest interval score. If the similarity is less than a similarity threshold, then the lower and upper limits of the height interval corresponding to the container with the highest interval score are expanded simultaneously, for example, by 0.1. Then the height interval after the first expansion is (-0.2, 0.1).

[0062] Furthermore, determining the similarity between the point cloud features in the target container and the point cloud features in the container with the highest interval score includes:

[0063] The point cloud points in the container with the highest interval score are determined as the first point cloud set, and the point cloud points in the target container are determined as the second point cloud set;

[0064] Determine the number of point cloud points in the second point cloud set whose distance from the first point cloud set is less than a second threshold.

[0065] The similarity is determined based on the number of point cloud points; the more point cloud points there are, the higher the similarity.

[0066] Specifically, when determining the distance between a point in the second point cloud set and the first point cloud set, it can be either determining the distance between a point in the second point cloud set and the center point of the first point cloud set, or determining the sum of the distances between a point in the second point cloud set and all points in the first point cloud set, and then determining the sum of these distances as the distance between the point in the second point cloud set and the first point cloud set.

[0067] By analyzing the distance between point clouds, we can identify whether the point clouds in other containers near the container with the highest interval score are continuously distributed with the point cloud in the container with the highest interval score. If they are continuously distributed, we expand the interval to find as many ground points as possible and accurately, thereby improving the accuracy of ground point segmentation.

[0068] It should be noted that the above steps describe how to segment a planar ROI to obtain the corresponding ground points. When there are multiple planar ROIs, repeating the above operations will obtain the ground points corresponding to each ROI.

[0069] Figure 3 This is a schematic diagram of a ground point cloud segmentation device according to an embodiment of the present disclosure. The ground point cloud segmentation device includes: a first determining module 310, used to determine a planar ROI based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information; a segmentation module 320, used to divide the same planar ROI into multiple containers according to different height intervals; a statistics module 330, used to count the point cloud points in the current frame point cloud that fall into each of the containers; a second determining module 340, used to determine the interval score of each container based on the point cloud points in each container, wherein the higher the interval score, the greater the probability that the point cloud point in the corresponding container is a ground point; and a third determining module 350, used to determine the ground point based on the container with the highest interval score.

[0070] Furthermore, the second determining module 340 includes a first determining unit, which is used to determine the interval score of each container for different height intervals on the same plane ROI, starting from the container with the lowest height interval, based on the total number of point cloud points in the container and a preset reward rule.

[0071] Furthermore, the first determining unit includes: a first determining subunit, used to determine the reward points of the current container based on the preset reward rule; and a second determining subunit, used to determine the sum of the total number of point cloud points in the current container and the reward points as the interval score of the current container.

[0072] Furthermore, the first determining subunit is specifically used to: determine that the current container receives reward points if the total number of point cloud points in the current container is greater than the interval score of the previous container adjacent to the current container, and / or if the range of continuous distribution of point cloud points in the current container reaches a first threshold; wherein the number of reward points is determined based on the total number of point cloud points in the current container, and / or based on the height interval corresponding to the current container, the more point cloud points in the current container, the more reward points; the higher the height interval corresponding to the current container, the fewer reward points.

[0073] Furthermore, the second determining module 340 is specifically used to: for the current container in each container, determine the interval score of the current container based on the total number of point cloud points in the current container, and / or the interval score of the containers adjacent to the current container, and / or the range of continuous distribution of point cloud points in the current container.

[0074] Furthermore, the third determining module 350 is specifically used to: determine whether the similarity between the point cloud features in the target container and the point cloud features in the container with the highest interval score is less than a similarity threshold, wherein the target container is a container adjacent to the container with the highest interval score; if the similarity is less than the similarity threshold and the number of target containers is one, determine whether the target container is above or below the container with the highest interval score.

[0075] If the target container is above the container with the highest interval score, the upper limit of the height interval corresponding to the container with the highest interval score is expanded. If the target container is below the container with the highest interval score, the lower limit of the height interval corresponding to the container with the highest interval score is expanded to obtain the height interval after the first expansion. If the similarity is less than the similarity threshold and there are two target containers, the upper and lower limits of the height interval corresponding to the container with the highest interval score are expanded simultaneously to obtain the height interval after the first expansion.

[0076] The lower and upper limits of the height range after the first expansion are expanded by a set value to obtain the height range after the second expansion.

[0077] Point cloud points falling within the second expanded altitude range are identified as ground points.

[0078] Determining the similarity between point cloud features in the target container and point cloud features in the container with the highest interval score includes:

[0079] The point cloud points in the container with the highest interval score are determined as the first point cloud set, and the point cloud points in the target container are determined as the second point cloud set;

[0080] Determine the number of point cloud points in the second point cloud set whose distance from the first point cloud set is less than a second threshold.

[0081] The similarity is determined based on the number of point cloud points; the more point cloud points there are, the higher the similarity.

[0082] The ground point cloud segmentation device provided in this embodiment can execute the steps in the ground point cloud segmentation method provided in this embodiment and obtain the same beneficial effects, which will not be repeated here.

[0083] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this disclosure. See below for details. Figure 4 It shows a schematic diagram of a structure suitable for implementing the electronic device 500 in the embodiments of this disclosure. Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0084] like Figure 4 As shown, the electronic device 500 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 501, which can perform various appropriate actions and processes to implement the methods of the embodiments described herein, based on a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing device 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0085] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts, thereby implementing the ground point cloud segmentation method as described above. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0086] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0087] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: determine a planar Region of Interest (ROI) based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information; divide the same planar ROI into multiple containers according to different height intervals; count the point cloud points in the current frame point cloud that fall into each of the containers; determine the interval score of each container based on the point cloud points in each container, wherein a higher interval score indicates a greater probability that the point cloud point in the corresponding container is a ground point; and determine the ground point based on the container with the highest interval score.

[0088] Optionally, when one or more of the above-described procedures are executed by the electronic device, the electronic device may also perform other steps described in the above embodiments.

[0089] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0090] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

Claims

1. A method for segmenting ground point clouds, characterized in that, The method includes: Determine the planar ROI based on the distribution characteristics of the current frame point cloud and / or the corresponding road network information; Divide the same planar ROI into multiple containers according to different height ranges; Count the point cloud points in the current frame that fall into each of the aforementioned containers; The interval score of each container is determined based on the point cloud points in each container. The higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points. The ground point is determined based on the container with the highest interval score, including: Determine whether the similarity between the point cloud features in the target container and the point cloud features in the container with the highest interval score is less than a similarity threshold, wherein the target container is the container adjacent to the container with the highest interval score; If the similarity is less than the similarity threshold, when there is only one target container, the height range corresponding to the container with the highest interval score is expanded upwards or downwards based on the relative position between the target container and the container with the highest interval score to obtain an expanded height range; if the similarity is less than the similarity threshold, when there are two target containers, the height range corresponding to the container with the highest interval score is expanded upwards and downwards to obtain an expanded height range. Point cloud points falling within the expanded altitude range are identified as ground points.

2. The method according to claim 1, characterized in that, The process of determining the interval score for each container based on the point cloud points in each container includes: For containers in different height ranges on the same plane ROI, starting from the container with the lowest height range, the range score of each container is determined based on the total number of point cloud points in the container and the preset reward rules.

3. The method according to claim 2, characterized in that, The determination of the interval score for each container based on the total number of point cloud points in the container and a preset reward rule includes: The reward points for the current container are determined based on the preset reward rules; The sum of the total number of point cloud points in the current container and the reward points is determined as the interval score of the current container.

4. The method according to claim 3, characterized in that, The step of determining the reward points for the current container based on the preset reward rule includes: If the total number of point cloud points in the current container is greater than the interval score of the previous container adjacent to the current container, and / or if the range of continuous distribution of point cloud points in the current container reaches a first threshold, then the current container is determined to receive reward points; wherein, the number of reward points is determined based on the total number of point cloud points in the current container, and / or based on the height interval corresponding to the current container, the more point cloud points in the current container, the more reward points; the higher the height interval corresponding to the current container, the fewer reward points.

5. The method according to claim 1, characterized in that, The process of determining the interval score for each container based on the point cloud points in each container includes: For the current container in each container, the interval score of the current container is determined based on the total number of point cloud points in the current container, and / or the interval score of the containers adjacent to the current container, and / or the range of continuous distribution of point cloud points in the current container.

6. The method according to claim 1, characterized in that, When the number of target containers is one, the height range corresponding to the container with the highest interval score is expanded upwards or downwards based on the relative position between the target container and the container with the highest interval score, to obtain the expanded height range, including: Determine whether the target container is above or below the container with the highest score in the interval; If the target container is above the container with the highest interval score, the height interval corresponding to the container with the highest interval score is expanded upwards; if the target container is below the container with the highest interval score, the height interval corresponding to the container with the highest interval score is expanded downwards, thus obtaining the height interval after the first expansion. When there are two target containers, the height interval corresponding to the container with the highest interval score is expanded upwards and downwards to obtain an expanded height interval, including: The upper and lower limits of the height interval corresponding to the container with the highest interval score are simultaneously expanded to obtain the height interval after the first expansion. The lower and upper limits of the height range after the first expansion are expanded by a set value to obtain the height range after the second expansion. The step of determining point cloud points falling within the expanded height range as ground points includes: Point cloud points falling within the second expanded altitude range are identified as ground points.

7. The method according to claim 6, characterized in that, Determining the similarity between point cloud features in the target container and point cloud features in the container with the highest interval score includes: The point cloud points in the container with the highest interval score are determined as the first point cloud set, and the point cloud points in the target container are determined as the second point cloud set; Determine the number of point cloud points in the second point cloud set whose distance from the first point cloud set is less than a second threshold. The similarity is determined based on the number of point cloud points; the more point cloud points there are, the higher the similarity.

8. A ground point cloud segmentation device, characterized in that, include: The first determining module is used to determine the planar ROI based on the distribution characteristics of the point cloud in the current frame and / or the corresponding road network information; The partitioning module is used to divide the same planar ROI into multiple containers according to different height ranges; The statistics module is used to count the point cloud points that fall into each of the containers in the current frame point cloud; The second determining module is used to determine the interval score of each container based on the point cloud points in each container. The higher the interval score, the greater the probability that the point cloud points in the corresponding container are ground points. The third determination module is used to determine the ground point based on the container with the highest interval score, including: Determine whether the similarity between the point cloud features in the target container and the point cloud features in the container with the highest interval score is less than a similarity threshold. The target container is the container adjacent to the container with the highest interval score. If the similarity is less than the similarity threshold, when there is only one target container, expand the height interval corresponding to the container with the highest interval score upwards or downwards according to the relative orientation between the target container and the container with the highest interval score to obtain an expanded height interval. If the similarity is less than the similarity threshold, when there are two target containers, expand the height interval corresponding to the container with the highest interval score both upwards and downwards to obtain an expanded height interval. Point cloud points falling within the expanded altitude range are identified as ground points.

9. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.

Images