A point cloud processing method of a laser radar, a laser radar and a storage medium

Abstract

This application provides a point cloud processing method for lidar, a lidar, and a storage medium. The method, applied in the lidar field, includes: acquiring point cloud data of the current frame collected by the lidar and the global reflectivity feature value of the previous frame; determining high-reflectivity regions and effective dilatation regions surrounding the high-reflectivity regions based on the point cloud data of the current frame; determining whether the effective dilatation regions surrounding the high-reflectivity regions are abnormal based on the global reflectivity feature value; and marking the high-reflectivity regions and the effective dilatation regions if the effective dilatation regions are determined to be abnormal. This method can accurately determine whether the current lidar point cloud is abnormal, avoiding the impact of abnormal point clouds on the perception accuracy of upper-layer sensing algorithms.

Description

Technical Field

[0001] This application relates to the field of lidar, and more specifically, to a point cloud processing method for lidar, lidar, and storage medium in the field of lidar. Background Technology

[0002] LiDAR is an active remote sensing device that uses a laser as the emission source and employs photoelectric detection technology. It is an advanced detection method that combines laser technology with modern photoelectric detection technology.

[0003] The relevant technologies cannot accurately determine whether the current point cloud of the LiDAR is abnormal. If the point cloud is abnormal, it will cause the upper-layer perception algorithm to misidentify, which may lead to frequent adjustments of the control algorithm's strategy due to misidentification, thus affecting the user experience. Summary of the Invention

[0004] This application provides a point cloud processing method for lidar, a lidar, and a storage medium. The method can accurately determine whether the current lidar point cloud is abnormal, thus avoiding the impact of abnormal point clouds on the perception accuracy of the upper-layer perception algorithm.

[0005] In a first aspect, a point cloud processing method for a lidar is provided. The method includes: acquiring point cloud data of the current frame collected by the lidar and the global reflectivity feature value of the previous frame of the current frame; determining a high-reflectivity region and an effective dilatation region surrounding the high-reflectivity region in the current frame based on the point cloud data of the current frame; determining whether the effective dilatation region surrounding the high-reflectivity region is abnormal based on the global reflectivity feature value; and marking the high-reflectivity region and the effective dilatation region if the effective dilatation region is determined to be abnormal.

[0006] The above technical solution establishes a temporal context by introducing the global reflectivity feature value of the previous frame. Considering that the difference between point cloud data in two consecutive frames is small under normal circumstances, using the global reflectivity feature value as a benchmark provides a reliable reference standard for judging whether there is abnormal dilation in the current frame. After identifying the high-reflectivity region and the effective dilation region around the high-reflectivity region in the current frame, the global reflectivity feature value can effectively distinguish abnormal point clouds caused by high-reflectivity dilation. Using the global reflectivity feature value of the previous frame, i.e., historical data, as a reference reduces misjudgments caused by fluctuations in data from a single frame and improves the accuracy of judging whether the effective dilation region is abnormal. After determining that the effective dilation region is abnormal, the high-reflectivity region and the effective dilation region are marked to avoid the upper-layer perception algorithm making misjudgments based on abnormal point cloud data, thereby reducing unnecessary adjustments to the control strategy. The marked point cloud data provides clear guidance for subsequent processing and helps the control algorithm make more accurate decisions.

[0007] In conjunction with the first aspect, in some possible implementations, determining the high-reflectivity region in the current frame based on the point cloud data of the current frame includes: determining whether a high-reflectivity sub-region exists in each scan sub-region based on the point cloud data of the current frame, thereby obtaining each high-reflectivity sub-region existing in the current frame; wherein, each high-reflectivity sub-region includes n consecutive high-reflectivity points, wherein the high-reflectivity points are scan points that meet preset high-reflectivity point requirements, and n is greater than a preset number; merging the high-reflectivity sub-regions in each high-reflectivity sub-region whose field-of-view overlap is greater than a preset overlap threshold, thereby obtaining the high-reflectivity region in the current frame.

[0008] In the above implementation, the point cloud data of the current frame is divided into multiple scanning sub-regions. Identification is performed using each scanning sub-region as a benchmark, ensuring that each high-reflectivity sub-region is precisely identified, improving the accuracy of local detail recognition and thus enhancing overall recognition precision. If there is significant overlap in the fields of view among the identified high-reflectivity sub-regions, these regions are merged into a larger high-reflectivity region, which better reflects the shape and size of the actual physical object, rather than dividing it into multiple small regions. Considering that in complex environments, different scanning angles may produce partially overlapping but not entirely consistent high-reflectivity sub-regions, merging these regions provides more consistent and reliable detection results. Therefore, the above implementation method of first identifying high-reflectivity sub-regions and then merging them not only better reflects the shape and size of the actual physical object but also reduces misidentification caused by local viewpoint differences, improving the accuracy of the final determined high-reflectivity region.

[0009] In combination with the first aspect and the above implementation methods, in some possible implementation methods, the point cloud data of the current frame includes: the distance measurement value and reflectance measurement value of each scan point; the preset high reflectance point requirement includes: the difference between the distance measurement values ​​of any two adjacent high reflectance points among the n high reflectance points is less than a first preset difference value, and the reflectance measurement values ​​of the n high reflectance points are all greater than a first preset reflectance threshold.

[0010] Combining the first aspect and the above implementation methods, in some possible implementation methods, the effective expansion region around the high-reflectivity region includes: an effective expansion region in a first direction around the high-reflectivity region, where the first direction is the scanning direction of the lidar; determining the effective expansion region around the high-reflectivity region based on the point cloud data of the current frame includes: for each high-reflectivity sub-region existing in the current frame, determining whether there is an expansion sub-region in the first direction around the high-reflectivity sub-region based on the point cloud data of the current frame, thereby obtaining the expansion sub-region corresponding to each high-reflectivity sub-region in the current frame; wherein, the scanning points in the expansion sub-region are all expansion points that meet the preset expansion point requirements; and determining the effective expansion region in the first direction around the high-reflectivity region based on the expansion sub-region corresponding to each high-reflectivity sub-region in the current frame.

[0011] In combination with the first aspect and the above implementation methods, in some possible implementation methods, the point cloud data of the current frame includes: the distance measurement value and reflectance measurement value of each scan point; the preset dilation point requirement includes: the difference between the distance measurement value of the dilation point and the distance measurement value of the high reflectance point in the high reflectance sub-region is less than a second preset difference value, and the reflectance measurement value of the dilation point is less than a second reflectance threshold value.

[0012] In combination with the first aspect and the above implementation methods, in some possible implementation methods, the step of determining whether the effective expansion region around the high reflectivity region is abnormal based on the global reflectivity feature value includes: determining a first expansion rate upper limit value of the high reflectivity sub-region based on the global reflectivity feature value; for each high reflectivity sub-region, calculating the expansion rate of the high reflectivity sub-region; if the expansion rate of the high reflectivity sub-region is greater than the first expansion rate upper limit value, then determining that the expansion sub-region corresponding to the high reflectivity sub-region is abnormal.

[0013] In combination with the first aspect and the above implementation methods, in some possible implementation methods, the calculation of the expansion rate of the high-reflectivity sub-region includes: determining the number of high-reflectivity points in the high-reflectivity sub-region and the number of expansion points in the corresponding expansion sub-region; and using the ratio obtained by dividing the number of expansion points by the number of high-reflectivity points as the expansion rate of the high-reflectivity sub-region.

[0014] Combining the first aspect and the above implementation, the effective expansion region around the high-reflectivity region includes: an effective expansion region in a second direction around the high-reflectivity region, where the second direction is perpendicular to the scanning direction of the lidar; determining the effective expansion region around the high-reflectivity region based on the point cloud data of the current frame includes: determining a reference expansion region located in the second direction around the high-reflectivity region based on the point cloud data of the current frame; wherein the boundary of the reference expansion region in the second direction is the same as the boundary of the high-reflectivity region, and the boundary of the reference expansion region in the scanning direction is determined based on the location of an expansion point that meets a preset expansion point requirement; identifying whether a real object exists within the reference expansion region; if a real object is identified, removing the area where the real object is located from the reference expansion region to obtain the effective expansion region in the second direction around the high-reflectivity region.

[0015] In the above implementation, by removing the region where the real object is located from the reference dilation region, it is beneficial to avoid marking real objects that should not be marked during subsequent labeling, thus providing accurate reference data for the upper-layer perception algorithm.

[0016] Combining the first aspect and the above implementation method, the step of identifying whether a real object exists in the reference expansion region includes: determining the number of theoretical scan points and the number of effective scan points in the reference expansion region; calculating the ratio of the number of effective scan points to the number of theoretical scan points; and if the ratio is greater than a preset ratio threshold, determining that a real object exists in the reference expansion region.

[0017] The above implementation incorporates the physical characteristics of LiDAR, namely that only real objects will trigger its effective scanning points, making the detection of real objects more accurate and reliable. The use of a preset ratio threshold ensures that only when the effective scanning points reach a certain density are they considered real objects, avoiding false alarms caused by isolated points.

[0018] Combining the first aspect and the above implementation method, the step of identifying whether a real object exists in the reference expansion region includes: determining the shape formed by each expansion point in the reference expansion region; determining whether the shape is a preset shape of a real object; if the shape is the preset shape, then determining that a real object exists in the reference expansion region.

[0019] In the above implementation method, by determining the shape of each expansion point in the reference expansion region, the spatial distribution characteristics of the expansion points can be understood, which helps to identify the real objects in the reference expansion region.

[0020] Combining the first aspect and the above implementation method, the step of determining whether the effective expansion region around the high-reflectivity region is abnormal based on the global reflectivity feature value includes: determining a second expansion rate upper limit value for the high-reflectivity region based on the global reflectivity feature value; calculating the expansion rate of the high-reflectivity region based on the number of high-reflectivity points within the high-reflectivity region and the number of expansion points in the effective expansion region in the second direction surrounding the high-reflectivity region; if the expansion rate of the high-reflectivity region is greater than the second expansion rate upper limit value, then determining that the effective expansion region in the second direction surrounding the high-reflectivity region is abnormal.

[0021] Combining the first aspect and the above implementation method, the step of determining whether the effective expansion region around the high reflectivity region is abnormal based on the global reflectivity feature value includes: determining a threshold for the number of expansion points in the high reflectivity region based on the global reflectivity feature value; determining the number of expansion points in the effective expansion region in the second direction around the high reflectivity region; if the number of expansion points in the effective expansion region in the second direction is greater than the threshold for the number of expansion points, then determining that the effective expansion region in the second direction around the high reflectivity region is abnormal.

[0022] Combining the first aspect and the above implementation method, when the current frame is not the first frame, the global reflectance feature value of the previous frame is calculated in the following way: scan points with reflectance measurement values ​​greater than the first reflectance threshold and scan points with reflectance measurement values ​​of 0 are removed from the point cloud data of the previous frame to obtain the remaining scan points; the average reflectance of the remaining scan points is calculated based on the reflectance measurement values ​​of the remaining scan points, and the average reflectance is used as the global reflectance feature value of the previous frame.

[0023] The above implementation method effectively removes data noise that may be caused by high reflectivity or other abnormalities by eliminating scan points with high reflectivity and zero reflectivity. The remaining scan points after filtering are closer to the real environmental reflectivity distribution, so that the global reflectivity feature value calculated based on the reflectivity measurement value of the remaining scan points can provide a more reliable basic reference for subsequent comparative analysis with the current frame point cloud data.

[0024] In a second aspect, a lidar is provided, including a memory for storing executable program code; and a processor for calling and running the executable program code from the memory, causing the lidar to perform the method described in the first aspect or any possible implementation thereof.

[0025] Thirdly, a computer program product is provided, comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof.

[0026] Fourthly, a computer-readable storage medium is provided that stores computer program code, which, when executed on a computer, causes the computer to perform the methods described in the first aspect or any possible implementation thereof. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the system composition of a lidar provided in an embodiment of this application;

[0028] Figure 2 This is a schematic flowchart of a point cloud processing method for a lidar provided in an embodiment of this application;

[0029] Figure 3 This is a schematic diagram of a partial point cloud scanned by a lidar according to an embodiment of this application;

[0030] Figure 4 This is a schematic diagram of a point cloud containing a low-reflection area of ​​a wall surface, provided in an embodiment of this application.

[0031] Figure 5 This is a schematic diagram of a point cloud containing a low-reflection pole region, provided in an embodiment of this application.

[0032] Figure 6 This is a schematic diagram illustrating two-dimensional marking of high-reflectivity regions and anomalous effective expansion regions provided in an embodiment of this application;

[0033] Figure 7 This is a high-inflection dilatation point cloud map for rainy weather provided in an embodiment of this application;

[0034] Figure 8 This application provides a labeled rainy day high-inflection expansion top-down point cloud map;

[0035] Figure 9 This application provides a marked high-inverse-expansion frontal point cloud map of rainy weather;

[0036] Figure 10 This is a flowchart illustrating another point cloud processing method for lidar provided in an embodiment of this application;

[0037] Figure 11 This is a schematic diagram of the structure of a lidar provided in an embodiment of this application. Detailed Implementation

[0038] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0039] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0040] LiDAR (Light Detection and Ranging) is an active remote sensing device that uses a laser as its emission source and employs photoelectric detection technology. It represents an advanced detection method that combines laser technology with modern photoelectric detection techniques. LiDAR's advantages include high resolution, high sensitivity, strong anti-interference capabilities, and immunity to darkness. It is widely used in autonomous driving, logistics vehicles, robotics, vehicle-to-everything (V2X) systems, and intelligent public transportation.

[0041] Figure 1 This is a schematic diagram of the system composition of a lidar provided in an embodiment of this application.

[0042] For example, such as Figure 1 The lidar system comprises a transmitting system 101, a receiving system 102, a scanning control system 103, a data processing system 104, and a point cloud output system 105. Its working principle is as follows: the transmitting system 101 transmits a detection signal to the target (illustratively shown as a vehicle in the diagram); the receiving system 102 receives the echo signal; the data processing system 104 processes the received echo signal to obtain information such as the target's distance, size, speed, and reflectivity; and the point cloud output system 105 outputs a point cloud.

[0043] Reflectance is a physical quantity that characterizes an object's optical reflectivity. Any object in nature, after being illuminated by light, will exhibit absorption and reflection of incident light. Different types of objects possess different electromagnetic wave properties, and therefore their reflected incident light properties also differ. In other words, given a fixed amount of incident light, the intensity of the reflected light will vary depending on the target object. For example, road signs are highly reflective objects; they have an extremely strong ability to reflect incident light. Even a small amount of weak energy light emanating from a highly reflective sign will be almost entirely reflected back along its original path, making it relatively easy to detect in a receiving system.

[0044] The reflectivity of a target object can usually be calculated based on the ratio of the received echo energy to the emitted energy. When the emitted beam is ideal, assuming it is a straight line, the beam enters the high-reflectivity sign from an open target without any transition during its rotation. Since the beam is an ideal straight line, there is no transition when it sweeps across the high-reflectivity sign, so the point cloud result represents the actual width of the high-reflectivity sign.

[0045] Because the actual beam emitted by a lidar is not an ideal beam, its spot has a certain area. When there is a highly reflective object next to the measured point, the diffuse reflection signal of the beam hitting the highly reflective object will affect the received echo at the current measured point. This causes the point cloud outline around the highly reflective object to spread outwards, forming a false point cloud with lower reflectivity, affecting the accuracy of ranging. This results in a high-reflectivity expansion phenomenon, which is particularly pronounced in rainy weather. This is mainly due to factors such as raindrops in the air and water adhering to the radar window, which enlarge the spot diameter and change the propagation path of the light. As more of the beam enters the high-reflectivity object, the echo energy gradually increases until the entire beam enters the high-reflectivity object, presenting a highly reflective form. The resulting point cloud is wider than the actual width of the high-reflectivity object, leading to inaccurate ranging. The point cloud outside the actual width of the high-reflectivity object presents a low-reflectivity form due to the low energy at the beam edge, while the point cloud within the actual width of the high-reflectivity object presents a high-reflectivity form due to the high energy of the main beam. This phenomenon is called high-reflectivity expansion.

[0046] The existing technology cannot accurately determine whether the current LiDAR point cloud is abnormal. When operating in rainy environments, especially in scenes with highly reflective objects, high-reflection inflation can easily occur, leading to misidentification by the upper-layer perception algorithm. This misidentification may then cause frequent adjustments to the policy control algorithm, impacting the user experience. Specifically, when high-reflection inflation occurs, the point cloud data output by the LiDAR may appear abnormal, such as false obstacle outlines or enlarged obstacle sizes. The upper-layer perception algorithm relies on this data for environmental understanding and obstacle detection, making it prone to misjudgment. When the perception algorithm incorrectly identifies an obstacle and deems it a threat to the vehicle, it triggers a response from the policy control algorithm, such as adjusting speed, changing lanes, or emergency braking. Such frequent and unnecessary policy adjustments not only affect driving smoothness but may also cause passenger discomfort and increase the risk of accidents.

[0047] For example, if a vehicle is equipped with LiDAR, during driving, the upper-level perception algorithm identifies obstacles based on the point cloud output by the LiDAR. If it deems an obstacle encroaching on the vehicle's drivable area, it will adjust the vehicle's control strategy, such as slowing down, changing lanes, or stopping. However, if the obstacle identified by the upper-level perception algorithm is a misidentification due to high inversion expansion, then these adjustments to the control strategy will be unnecessary and may even negatively impact driving safety and passenger experience.

[0048] Therefore, accurately determining whether the current point cloud of a LiDAR is abnormal has become an urgent technical problem to be solved. To at least solve the above-mentioned technical problem, embodiments of this application provide a point cloud processing method for LiDAR, which can accurately determine whether the current point cloud of the LiDAR is abnormal, and avoid abnormal point clouds affecting the perception accuracy of upper-layer perception algorithms.

[0049] Figure 2 This is a schematic flowchart of a point cloud processing method for lidar provided in an embodiment of this application.

[0050] For example, such as Figure 2 As shown, the point cloud processing method of this lidar includes the following steps 201 to 204:

[0051] Step 201: Obtain the point cloud data of the current frame collected by the lidar and the global reflectivity feature value of the previous frame.

[0052] Step 202: Based on the point cloud data of the current frame, determine the high-reflectivity region in the current frame and the effective dilatation region around the high-reflectivity region.

[0053] Step 203: Based on the global reflectivity characteristic value, determine whether the effective expansion region around the high reflectivity region is abnormal.

[0054] Step 204: If the effective expansion region is determined to be abnormal, then mark the high-reflection region and the effective expansion region.

[0055] exist Figure 2 In the illustrated embodiment, by introducing the global reflectivity feature value of the previous frame, a correlation of detection data in the time dimension is established. Considering that the difference between point cloud data in two consecutive frames is small under normal circumstances, the global reflectivity feature value is used as a benchmark to provide a reliable reference standard for determining whether there is abnormal dilation in the current frame. After determining the high reflectivity region and the effective dilation region around the high reflectivity region in the current frame, the global reflectivity feature value can be used to effectively distinguish abnormal point clouds caused by high reflectivity dilation. Using the global reflectivity feature value of the previous frame, i.e., historical data, as a reference reduces misjudgments caused by fluctuations in data from a single frame and improves the accuracy of determining whether the effective dilation region is abnormal. After determining that the effective dilation region is abnormal, the high reflectivity region and the effective dilation region are marked to avoid the upper-layer perception algorithm making misjudgments based on abnormal point cloud data, thereby reducing unnecessary adjustments to the control strategy. The marked point cloud data provides clear guidance for subsequent processing and helps the control algorithm make more accurate decisions.

[0056] The following section uses the application of point cloud processing methods for LiDAR to LiDAR as an example to explain in detail the implementation of steps 201 to 204 above:

[0057] In step 201, the current frame point cloud data acquired by the lidar includes three-dimensional spatial information of the environment (distance, position) and reflectivity information (reflection intensity). The global reflectivity feature value is calculated based on the reflectivity of the previous frame point cloud data, reflecting the average reflectivity of the previous frame point cloud data.

[0058] If the current frame is the first frame, the pre-set default global reflectivity feature value can be used as the global reflectivity feature value of the previous frame.

[0059] When the current frame is not the first frame, the calculation method for the global reflectivity feature value of the previous frame can include the following steps S11 to S12:

[0060] S11: Remove scan points with reflectance measurements greater than the first reflectance threshold and scan points with reflectance measurements of 0 from the point cloud data of the previous frame to obtain the remaining scan points.

[0061] The first reflectivity threshold can be pre-calibrated to measure whether the reflectivity measurement value of the scanned point is too high. The point cloud data of the previous frame includes the position information (x, y, z coordinates) and reflectivity measurement value of each scanned point. The process iterates through all scanned points in the previous frame, checking the reflectivity measurement value of each point. If the reflectivity measurement value of a scanned point is greater than the set first reflectivity threshold, it indicates that the reflectivity measurement value of that scanned point is too high, and it is removed from the point cloud data. If the reflectivity measurement value of a scanned point is equal to 0, it is also removed from the point cloud data, resulting in the remaining scanned points. The remaining scanned points can be understood as scanned points with normal reflectivity, while scanned points with reflectivity measurement values ​​greater than the first reflectivity threshold and scanned points with reflectivity measurement values ​​of 0 can be understood as scanned points with abnormal reflectivity.

[0062] S12: Calculate the average reflectance of the remaining scan points based on the reflectance measurement values ​​of the remaining scan points, and use the average reflectance as the global reflectance feature value of the previous frame.

[0063] Specifically, the reflectance measurements of all remaining scan points are summed and then divided by the number of remaining scan points to obtain the average reflectance. The calculated average reflectance is used as the global reflectance feature value of the previous frame for subsequent comparative analysis with the point cloud data of the current frame.

[0064] The above implementation method effectively removes data noise that may be caused by high reflectivity or other abnormalities by eliminating scan points with high reflectivity and zero reflectivity. The remaining scan points after filtering are closer to the real environmental reflectivity distribution, so that the global reflectivity feature value calculated based on the reflectivity measurement value of the remaining scan points can provide a more reliable basic reference for subsequent comparative analysis with the current frame point cloud data.

[0065] In the specific implementation, for each frame of point cloud data, the global reflectivity feature value can be calculated and stored as reference data to determine whether the dilated region in the next frame of point cloud is abnormal.

[0066] In step 202, based on the point cloud data of the current frame, high-reflectivity regions and effective dilatation regions surrounding the high-reflectivity regions are identified. For example, based on reflectivity measurements in the point cloud data, regions with reflectivity measurements significantly higher than a threshold are identified; these regions typically correspond to high-reflectivity objects. The effective dilatation region is an extended region surrounding the high-reflectivity region, used to capture anomalous point cloud data that may be caused by high-reflectivity dilatation.

[0067] For example, the implementation of determining the high-reflectivity region in the current frame based on the point cloud data of the current frame includes the following steps S21 to S22:

[0068] S21: Based on the point cloud data of the current frame, determine whether there is a high-reflectivity sub-region in each scanned sub-region, and obtain the high-reflectivity sub-regions that exist in the current frame.

[0069] The high-reflection sub-region includes n consecutive high-reflection points. A high-reflection point is a scan point that meets the preset high-reflection point requirements. n is greater than the preset number, which avoids isolated or scattered high-reflection points being mistakenly identified as high-reflection sub-regions.

[0070] In practical implementation, if the LiDAR scans by row, each scanned sub-region can be understood as a row in the point cloud data, and the corresponding high-reflectivity sub-region can be understood as a high-reflectivity row. If the LiDAR scans by column, each scanned sub-region can be understood as a column in the point cloud data, and the corresponding high-reflectivity sub-region can be understood as a high-reflectivity column. If the LiDAR scans by block, each scanned sub-region can be understood as a block in the point cloud data, and the corresponding high-reflectivity sub-region can be understood as a high-reflectivity block.

[0071] For example, the point cloud data of the current frame includes: the distance measurement value and reflectance measurement value of each scan point; the above-mentioned preset high reflectance point requirements include: the difference between the distance measurement values ​​of any two adjacent high reflectance points among the n high reflectance points is less than a first preset difference value, and the reflectance measurement values ​​of the n high reflectance points are all greater than a first preset reflectance threshold.

[0072] The first preset difference can be pre-calibrated to measure whether the difference in distance measurements between any two high-reflectivity points is small. The first preset reflectivity threshold can also be pre-calibrated to measure whether the reflectivity measurement of the scanned point is high enough. If the reflectivity measurement of a scanned point is greater than the first preset reflectivity threshold, it indicates that the scanned point is a high-reflectivity point. If the difference in distance measurements between any two adjacent high-reflectivity points among the n high-reflectivity points is less than the first preset difference, it indicates that the distance between any two adjacent high-reflectivity points among the n high-reflectivity points is relatively close, meaning that these n high-reflectivity points can form a high-reflectivity sub-region.

[0073] For example, if the LiDAR scans by line, then for each line in the point cloud data, it can be determined whether there are high-inflection lines in that line according to the method in S21 above, and then all high-inflection lines in the current frame can be obtained.

[0074] S22: Merge the high-reflectivity sub-regions in each high-reflectivity sub-region where the field of view overlap is greater than the preset overlap threshold to obtain the high-reflectivity region in the current frame.

[0075] The preset overlap threshold can be pre-calibrated to measure whether each high-reflectivity sub-region can be merged. If the field-of-view overlap between high-reflectivity sub-region 1 and high-reflectivity sub-region 2 is greater than the preset overlap threshold, then high-reflectivity sub-region 1 and high-reflectivity sub-region 2 are considered to be the same high-reflectivity region, and high-reflectivity sub-region 1 and high-reflectivity sub-region 2 can be merged.

[0076] For example, if each high-reflectivity sub-region is a high-reflectivity row, then the degree of overlap between the fields of view of each high-reflectivity row can be: the degree of overlap of the columns occupied by each high-reflectivity row. The degree of overlap of the columns occupied by any two high-reflectivity rows = the number of overlapping columns / the maximum number of columns, where the maximum number of columns is the number of columns occupied by the high-reflectivity row with the most columns. For example, the columns occupied by high-reflectivity row 1 include columns 5 to 20, and the columns occupied by high-reflectivity row 2 include columns 5 to 19. The number of overlapping columns between high-reflectivity rows 1 and 2 is 15. The total number of columns occupied by high-reflectivity row 1 is 16, and the total number of columns occupied by high-reflectivity row 2 is 15. Therefore, the degree of overlap between high-reflectivity rows 1 and 2 = 15 / 16.

[0077] For example, if each high-reflectivity sub-region is a high-reflectivity column, then the degree of overlap between the fields of view of each high-reflectivity column can be: the degree of overlap of the rows occupied by each high-reflectivity column. The degree of overlap of the rows occupied by any two high-reflectivity columns = the number of overlapping rows / the maximum number of rows, where the maximum number of rows is the number of rows occupied by the high-reflectivity column with the most rows. For example, the rows occupied by high-reflectivity column 3 include rows 5 to 20, and the rows occupied by high-reflectivity column 4 include rows 5 to 19. The number of overlapping rows between high-reflectivity columns 3 and 4 is 15. The total number of rows occupied by high-reflectivity column 3 is 16, and the total number of rows occupied by high-reflectivity column 4 is 15. Therefore, the degree of overlap between high-reflectivity columns 3 and 4 is 15 / 16.

[0078] For example, if each high-reflectivity sub-region is a high-reflectivity block, then the degree of overlap between the fields of view of each high-reflectivity block can be: the degree of overlap of the pixels occupied by each high-reflectivity block. The degree of overlap of the pixels occupied by any two high-reflectivity blocks = the number of overlapping pixels / the maximum number of pixels in the block, where the maximum number of pixels in the block can be the number of pixels corresponding to the block with the most pixels among two adjacent blocks.

[0079] In the above implementation, the point cloud data of the current frame is divided into multiple scanning sub-regions. Identification is performed using each scanning sub-region as a benchmark, ensuring that each high-reflectivity sub-region is precisely identified, thus improving recognition accuracy and overall precision. If there is significant overlap in the fields of view among the identified high-reflectivity sub-regions, these regions are merged into a larger high-reflectivity region, which better reflects the shape and size of the actual physical object, rather than dividing it into multiple small regions. Considering that in complex environments, different scanning angles may produce partially overlapping but not entirely consistent high-reflectivity sub-regions, merging these regions provides more consistent and reliable detection results. Therefore, the above implementation method of first identifying high-reflectivity sub-regions and then merging them not only better reflects the shape and size of the actual physical object but also reduces misidentification caused by local viewpoint differences, improving the accuracy of the final determined high-reflectivity region.

[0080] For example, considering that in real-world environments, LiDAR scanning is not always ideal, and various factors (such as environmental noise, changes in object surface characteristics, etc.) may cause discontinuities or anomalies in the point cloud data, a preset redundancy amount (also known as a fault tolerance space) can be introduced to improve the accuracy and robustness of high-reflectivity region identification. This allows for the existence of a preset number of non-high-reflectivity sub-regions within a high-reflectivity region without immediately terminating the expansion of the high-reflectivity region. For ease of understanding, the following explanation uses a high-reflectivity sub-region as an example of high-reflectivity behavior:

[0081] First, based on preset high-inflection point requirements, the high-inflection lines in the current frame are identified. For example, assume that lines 3 through 8 have been determined to be high-inflection lines within the high-inflection region. A preset redundancy limit (e.g., 2 lines) is set, meaning that a maximum of 2 consecutive non-high-inflection lines are allowed within the high-inflection region without immediately terminating the expansion of the high-inflection region. Starting from line 9, each line is checked for high-inflection. If line 9 is not a high-inflection line, line 10 is checked. If line 10 is also not a high-inflection line, the preset redundancy limit (2 lines) has been reached, and the high-inflection region ends here. If high-inflection lines are detected again within the redundancy limit (e.g., within 2 lines), these lines are included in the high-inflection region, and the region continues to expand. If the number of consecutive non-high-inflection lines reaches the preset redundancy limit, the high-inflection region is considered to have ended.

[0082] In the above implementation, by introducing a preset amount of redundancy (fault tolerance space), a certain number of non-high-reflectivity rows can be allowed during the identification of high-reflectivity regions. This avoids the entire high-reflectivity region being erroneously terminated due to misjudgment of a single or a few rows, thereby improving the tolerance for suboptimal scanning results. This method not only enhances the robustness and adaptability of the system but also reduces the possibility of missegmentation, ensuring the accuracy and stability of high-reflectivity region identification.

[0083] For example, the effective expansion region around the high-reflectivity region includes: an effective expansion region around the high-reflectivity region in a first direction, where the first direction is the scanning direction of the lidar. The determination of the effective expansion region around the high-reflectivity region based on the point cloud data of the current frame includes the following steps S31 to S32:

[0084] S31: For each high-reflectivity sub-region in the current frame, based on the point cloud data of the current frame, determine whether there is an expanded sub-region in the first direction around the high-reflectivity sub-region, and obtain the expanded sub-region corresponding to each high-reflectivity sub-region in the current frame; wherein, the scanning points in the expanded sub-region are all expanded points that meet the preset expansion point requirements.

[0085] As mentioned above, the scanning method of a lidar can be row-based scanning (scanning horizontally), column-based scanning (scanning vertically), or block-based scanning (scanning in a polling pattern according to an arbitrary set shape). Therefore, the first direction can be horizontal, vertical, or the direction of polling. When the first direction is horizontal, the effective expansion region around the high-reflection area includes the expansion sub-regions on the left and right sides of the high-reflection row. When the first direction is vertical, the effective expansion region around the high-reflection area includes the expansion sub-regions on the top and bottom sides of the high-reflection column. When the first direction is the polling direction, the expansion region consists of the expansion sub-regions on both sides perpendicular to the polling direction.

[0086] In some embodiments, the first direction is the horizontal direction, and the high-reflection sub-region is the high-reflection line. For each high-reflection line in the current frame, the scanning points on the left and right sides of the high-reflection line are checked, and these scanning points are examined one by one to see if they are expansion points that meet the preset expansion point requirements. The region formed by the expansion points on the left and right sides of the high-reflection line is the expansion sub-region.

[0087] In other embodiments, the first direction is the vertical direction, the high-inflection sub-region is the high-inflection column, for each high-inflection column in the current frame, the scanning points on the upper and lower sides of the high-inflection column are checked, and these scanning points are examined one by one to see if they are expansion points that meet the preset expansion point requirements. The area formed by the expansion points on the upper and lower sides of the high-inflection column is the expansion sub-region.

[0088] For example, the point cloud data of the current frame includes: distance measurement and reflectance measurement for each scan point. The preset dilation point requirements include: the difference between the distance measurement of the dilation point and the distance measurement of a high-reflectance point in the high-reflectance sub-region is less than a second preset difference, and the reflectance measurement of the dilation point is less than a second reflectance threshold.

[0089] The second preset difference can be pre-calibrated to characterize the proximity between the expansion point and other high-reflectivity points in the high-reflectivity sub-region. The second reflectivity threshold can also be pre-calibrated to characterize a smaller reflectivity measurement value at the expansion point. The preset expansion point requirements include a distance condition (the difference in distance measurements is less than the second preset difference) and a reflectivity condition (the reflectivity measurement value is less than the second reflectivity threshold). This is achieved by checking whether expansion points caused by the high-reflectivity expansion phenomenon in the high-reflectivity sub-region exist in the first direction surrounding the high-reflectivity sub-region from both distance and reflectivity dimensions. It is understood that there may be multiple high-reflectivity points in the high-reflectivity sub-region, and the distance condition in the preset expansion point requirements can be: the difference between the distance measurement value of the expansion point and the distance measurement value of any high-reflectivity point in the high-reflectivity sub-region is less than the second preset difference.

[0090] Figure 3 This is a schematic diagram of a partial point cloud scanned by a lidar according to an embodiment of this application.

[0091] For example, such as Figure 3 As shown, the first direction is the horizontal direction. Figure 3 This diagram illustrates the point cloud data of a single row obtained during a line-by-line scan by a lidar system. For Figure 3In the high-reflection row 301 (i.e., the high-reflection sub-region), it is determined whether each scan point in the left region 302 of the high-reflection row 301 meets the preset expansion point requirement, and also whether each scan point in the right region 303 of the high-reflection row 301 meets the preset expansion point requirement. If both scan points in the left region 302 meet the preset expansion point requirement, it is determined that an expansion sub-region exists on the left side of the high-reflection row 301. If both scan points in the right region 303 meet the preset expansion point requirement, it is determined that an expansion sub-region exists on the right side of the high-reflection row 301.

[0092] S32: Determine the effective expansion region in the first direction around each high-reflectivity sub-region in the current frame based on the expansion sub-regions corresponding to each high-reflectivity sub-region.

[0093] Specifically, the dilated sub-regions corresponding to each high-reflectivity sub-region in the current frame can all be considered as effective dilated regions in the first direction surrounding the high-reflectivity region. For example, if the first direction is horizontal, the dilated sub-regions on both sides of each high-reflectivity row can be considered as effective dilated regions. If the first direction is vertical, the dilated sub-regions on both sides of each high-reflectivity column can be considered as effective dilated regions.

[0094] In step 203, based on the global reflectance feature value of the previous frame, it is determined whether the effective dilatation region around the high reflectance region in the current frame is abnormal. Considering that under normal circumstances, the difference between the point cloud data of the two frames is small, the global reflectance feature value of the previous frame can provide a valid reference for the anomaly judgment of the current frame.

[0095] For example, step 203 above can be implemented by the following steps S41 to S42:

[0096] S41: Determine the upper limit of the first expansion rate of the high-reflectivity sub-region based on the global reflectivity characteristic value.

[0097] There is a pre-defined first correspondence between the global reflectivity feature value and the first expansion rate upper limit value. By querying this first correspondence, the first expansion rate upper limit value corresponding to the global reflectivity feature value can be obtained. The first expansion rate upper limit value represents the theoretical maximum expansion rate of the high reflectivity sub-region in the current frame when the expansion sub-region corresponding to the high reflectivity sub-region is normal, under the global reflectivity feature value condition of the previous frame.

[0098] In some embodiments, a target interval containing the global reflectivity feature value of the previous frame can be determined from multiple preset intervals. A first correspondence relationship is then queried based on the target interval to obtain the upper limit value of the first dilation rate corresponding to the target interval. The first correspondence relationship includes the correspondence between preset intervals and the upper limit value of the first dilation rate; the larger the upper limit value of the preset interval, the smaller the corresponding upper limit value of the first dilation rate.

[0099] For example, the global reflectivity feature value is specifically the global reflectivity mean (avg_ref_ratio), which is the reflectivity mean calculated based on the reflectivity measurements of the remaining scan points. In the case where the high-reflectivity sub-region is specifically the high-reflectivity row mentioned above, the first expansion ratio upper limit is for the high-reflectivity row; therefore, the first expansion ratio upper limit can be called the row expansion ratio upper limit. Based on this, the first correspondence can be as shown in Table 1 below:

[0100] Table 1

[0101] Global mean reflectance Upper limit of line expansion rate 0<avg_ref_ratio<=8 5.0 8<avg_ref_ratio<=14 3.0 14<avg_ref_ratio<=20 0.8 20<avg_ref_ratio<=26 0.5 26<avg_ref_ratio<255 0.01

[0102] Referring to Table 1, if the target interval of the mean global reflectance of the previous frame is (0, 8), then the upper limit of the expansion rate of the high reflectance region is 5.

[0103] S42: For each high-reflectivity sub-region, calculate the expansion rate of the high-reflectivity sub-region. If the expansion rate of the high-reflectivity sub-region is greater than the upper limit of the first expansion rate, then determine that the expansion sub-region corresponding to the high-reflectivity sub-region is abnormal.

[0104] For example, calculating the expansion rate of the high-reflectivity sub-region includes: determining the number of high-reflectivity points in the high-reflectivity sub-region and the number of expansion points in the corresponding expansion sub-region; and using the ratio obtained by dividing the number of expansion points by the number of high-reflectivity points as the expansion rate of the high-reflectivity sub-region.

[0105] Once the highly reflective sub-region and its corresponding dilated sub-region are determined, the number of highly reflective points in the highly reflective sub-region and the number of dilated points in the dilated sub-region can be determined. Taking a highly reflective sub-region as an example of highly reflective behavior, see [link to relevant documentation]. Figure 3 The number of high-reverse points in high-reverse row 301 is 10, and the number of expansion points in the expansion sub-regions (left region 301 and right region 303) corresponding to high-reverse row 301 is 4. Therefore, the expansion rate of high-reverse row 301 is 4 / 10 = 0.4.

[0106] For each high-inflectivity sub-region in the current frame, the dilation rate of the high-inflectivity sub-region is calculated and compared with the first upper limit value of dilation rate. If the dilation rate of the high-inflectivity sub-region is greater than the first upper limit value of dilation rate, the dilation sub-region corresponding to the high-inflectivity sub-region is determined to be abnormal. If the dilation rate of the high-inflectivity sub-region is less than or equal to the first upper limit value of dilation rate, the dilation sub-region corresponding to the high-inflectivity sub-region is determined to be normal.

[0107] For example, assuming the average global reflectance of the previous frame is 23, then according to Table 1 above, the upper limit of the line dilation rate corresponding to 23 is 0.5. Combining this with the above, the dilation rate of high-reflectivity line 301 is 0.4 < 0.5, therefore the dilated sub-region corresponding to high-reflectivity line 301 is determined to be normal. If the dilation rate of a certain high-reflectivity line in the current frame is greater than 0.5, then the dilated sub-region corresponding to that high-reflectivity line is determined to be abnormal.

[0108] For example, the effective expansion region around the high-reflectivity region includes: an effective expansion region in a second direction around the high-reflectivity region, where the second direction is perpendicular to the scanning direction of the lidar; the determination of the effective expansion region around the high-reflectivity region based on the point cloud data of the current frame includes the following steps S51 to S53:

[0109] S51: Based on the point cloud data of the current frame, determine the reference expansion region located in the second direction around the high-reflectivity region; wherein, the boundary of the reference expansion region in the second direction is the same as the boundary of the high-reflectivity region, and the boundary of the reference expansion region in the scanning direction is determined based on the location of the expansion point that meets the preset expansion point requirements.

[0110] As discussed above, if the scanning direction of the lidar is the first direction, then the second direction is the direction perpendicular to the first direction. When the second direction is horizontal, the reference expansion region can be located above and / or below the high-reflectivity region. When the second direction is vertical, the reference expansion region can be located to the left and / or right of the high-reflectivity region.

[0111] For example, if the scanning direction is horizontal, the boundary of the reference expansion region below the high-reflectivity region includes: the initial row and the ending row where the expansion point meeting the preset expansion point requirements is located below the high-reflectivity region. The initial row is the upper boundary of the reference expansion region, and the ending row is the lower boundary of the reference expansion region. The upper boundary of the reference expansion region is the row below the high-reflectivity region. For example, if the last row of the high-reflectivity region is row 10, then the upper boundary of the reference expansion region is row 11.

[0112] For example, a preset boundary limit can be set. If the ending row containing the expansion point below the high-reflectivity region does not exceed the preset boundary limit, then the ending row is used as the lower boundary of the reference expansion region. If the ending row exceeds the preset boundary limit, then the preset boundary limit is used as the lower boundary of the reference expansion region. Assuming the LiDAR point cloud has 126 rows, the preset boundary limit can be set to 72 rows. If the ending row containing the expansion point below the high-reflectivity region is row 60, then row 60 is used as the lower boundary of the reference expansion region in the scanning direction. If the ending row containing the expansion point below the high-reflectivity region exceeds row 72 (for example, row 73 contains expansion points), then row 72 is used as the lower boundary of the reference expansion region.

[0113] S52: Identify the presence of a real object within the reference expansion region.

[0114] Considering that some high-reflectivity areas are located on wall surfaces, or that there are low-reflectivity real objects below high-reflectivity areas, such as thin rods, it is necessary to identify whether real objects exist within the reference expansion area to avoid marking real objects that should not be marked during subsequent labeling.

[0115] In one possible implementation, the above-mentioned S52 includes the following S521 to S523:

[0116] S521: Determine the number of theoretical scan points and the number of effective scan points within the reference expansion region.

[0117] The theoretical scan point is the total number of scan points within the reference expansion region. Specifically, it refers to the number of points that the lidar should ideally scan within the reference expansion region. The ideal situation assumes that all emitted laser beams are reflected by objects and return to the receiver without any interference or loss. The number of theoretical scan points can be determined based on the row and column numbers of the four boundaries of the reference expansion region. For example, when the second direction is vertical, the number of theoretical scan points = (column number of the right boundary of the high-reflectivity region - column number of the left boundary of the high-reflectivity region) × number of rows in the reference expansion region.

[0118] The number of effective scan points refers to the number of echo points actually received by the lidar within the reference expansion region. These echo points indicate that the laser beam was successfully reflected back from the target object and detected by the lidar receiver. Effective scan points can be any scan points with a non-zero distance measurement value. The number of effective scan points is obtained by counting the number of scan points with non-zero distance measurement values ​​within the reference expansion region.

[0119] The theoretical scan points within the reference expansion region can include: valid scan points and invalid scan points. Invalid scan points are scan points with a distance measurement value of 0, while valid scan points are scan points with a distance measurement value of non-zero. Valid scan points can also be understood as expansion points within the reference expansion region.

[0120] S522: Calculate the ratio of the number of valid scan points to the number of theoretical scan points.

[0121] S523: If the above ratio is greater than the preset ratio threshold, then it is determined that there is a real object in the reference expansion region.

[0122] The preset ratio threshold can be pre-calibrated to measure whether the ratio of the number of effective scan points to the number of theoretical scan points is large enough. If the ratio is large enough, it indicates that there are more effective scan points, meaning that the effective scan points within the reference expansion region are sufficiently dense, and thus the probability of the presence of a real object is greater. Based on this, if the ratio of the number of effective scan points to the number of theoretical scan points is greater than the preset ratio threshold, it is determined that a real object exists within the reference expansion region; if the ratio is less than or equal to the preset ratio threshold, it is determined that no real object exists within the reference expansion region.

[0123] Figure 4 This is a schematic diagram of a point cloud containing a low-reflection area of ​​a wall surface, provided in an embodiment of this application.

[0124] For example, such as Figure 4 As shown, the scanning direction of the lidar is horizontal, and the second direction is perpendicular to the scanning direction. Therefore, the second direction is vertical, and the reference expansion region 402 surrounding the high-reflectivity region 401 is located below the high-reflectivity region 401. The 6th and 7th rows in the high-reflectivity region 401 are preset fault tolerance spaces, i.e., the preset redundancy is 2 rows.

[0125] The boundary of the reference expansion region 402 in the vertical direction is the same as the boundary of the high-reflection region 401. For example... Figure 4 As shown, the left boundary of the high-reflection region 401 in the vertical direction is the 3rd column, and the right boundary of the high-reflection region 401 in the vertical direction is the 10th column. Therefore, the left boundary of the reference expansion region 402 in the vertical direction is the 3rd column, and the right boundary is the 10th column. The upper boundary of the reference expansion region 402 in the horizontal direction is the initial row below the high-reflection region 401 where the expansion point that meets the preset expansion point requirements is located. Figure 4 In the diagram, the initial row containing the expansion point below the high-reflectivity region 401 is row 8; therefore, the upper boundary of the reference expansion region 402 is row 8. The lower boundary of the reference expansion region 402 in the horizontal direction is determined based on the ending row containing the expansion point below the high-reflectivity region 401 that meets the preset expansion point requirements. Figure 4 In the diagram, the line ending at the expansion point below the high-reflection region 401 is line 11. Assuming the preset boundary limit is line 72, since line 11 does not exceed line 72, line 11 is taken as the lower boundary of the reference expansion region 402. At this point, all four boundaries of the reference expansion region 402 are determined: the upper boundary is line 8, the lower boundary is line 12, the left boundary is column 3, and the right boundary is column 10. Therefore, the reference expansion region 402 can be determined based on these four boundaries.

[0126] The number of theoretical scan points within the reference expansion region 402 is = (column number of the right boundary of the high-reflectivity region - column number of the left boundary of the high-reflectivity region) × number of rows in the reference expansion region = (10-3) × 4 = 28. The number of invalid scan points (illustratively shown as solid black dots in the diagram) with a distance measurement value of 0 within the reference expansion region 402 is 9. The number of valid scan points within the reference expansion region 402 is = 28 - 9 = 19. Based on this, the ratio of valid scan points to theoretical scan points is = 19 / 28. Assuming the preset ratio threshold is set to 17 / 28, since 19 / 28 is greater than 17 / 28, it is determined that a real object exists in the reference expansion region 402. This real object could be, for example, a low-reflectivity wall.

[0127] The above implementation incorporates the physical characteristics of LiDAR, namely that only real objects will trigger its effective scanning points, making the detection of real objects more accurate and reliable. The use of a preset ratio threshold ensures that only when the effective scanning points reach a certain density are they considered real objects, avoiding false alarms caused by isolated points.

[0128] In another possible implementation, the above-mentioned S52 is implemented as follows: S524 to S526:

[0129] S524: Determine the shape formed by the expansion points within the reference expansion region.

[0130] Among them, the expansion point is the scan point in the reference expansion area that meets the above-mentioned preset expansion point requirements. After filtering out all the expansion points in the reference expansion area, a geometric fitting algorithm is used to model all the expansion points in order to determine the shape composed of each expansion point.

[0131] S525: Determine whether the shape is a preset shape of a real object.

[0132] S526: If the shape is a preset shape, then it is determined that there is a real object within the reference expansion area.

[0133] In this context, "real objects" refers to actual objects existing in the environment, such as poles, roadblocks, and walls. The preset shape can be any shape of a real object, such as the shape of a pole, roadblock, or wall. If the shape formed by the valid scan points within the reference expansion region matches the preset shape of a real object, then a real object exists within the reference expansion region. If the shape formed by the valid scan points does not match the preset shape of a real object, then a real object does not exist within the reference expansion region.

[0134] Figure 5 This is a schematic diagram of a point cloud containing a low-reflection pole region, provided in an embodiment of this application.

[0135] For example, such as Figure 5 As shown, the scanning direction of the lidar is horizontal, and the second direction is perpendicular to the scanning direction. Therefore, the second direction is vertical, and the reference expansion region 502 surrounding the high-reflectivity region 501 is located below the high-reflectivity region 501. The 6th and 7th rows in the high-reflectivity region 501 are a preset fault tolerance space, i.e., a preset redundancy of 2 rows. The reference expansion region 502 is related to the above... Figure 4 The method for determining the reference expansion region 402 is similar, and will not be repeated here to avoid repetition.

[0136] exist Figure 5 In the reference expansion region 502, the upper boundary is row 8, the lower boundary is row 12, the left boundary is column 3, and the right boundary is column 10. Considering that the shape characteristics of a real object like a rod in the point cloud are such that certain columns occupy a large number of low-reflection points, far exceeding the number of high-reflection points, histogram statistical analysis can be used to determine whether the shape formed by each effective scan point within the reference expansion region is the shape of a rod. For example, the number of expansion points in the reference expansion region 502 can be used as a statistical measure to construct a histogram. Figure 5 The histogram shown. In Figure 5 In the histogram shown, the number of expansion points in columns 3, 4, and 10 of the reference expansion region 502 is 0, the number of expansion points in columns 5 and 6 is 5, the number of expansion points in column 7 is 4, and the number of expansion points in columns 8 and 9 is 1. Then, a difference operation is performed on the histogram to obtain a difference array. Each value in the difference array is the number of expansion points in the next column minus the number of expansion points in the previous column. For example, 5 in the difference array is the difference between the number of expansion points in column 5 (5) and the number of expansion points in column 4 (0). Next, each value in the difference array is traversed point by point. When the traversed value A(5) is greater than the preset value pole_size_thr (for example, half of the number of rows in the reference expansion region 502, 5 / 2), the traversal continues. If the value B(-3) is traversed within the preset window length, and the value B(-3) has the opposite sign to the value A(5) (value A is positive and value B is negative) and the absolute value of value B is greater than the above preset value pole_size_thr(5 / 2), then the shape formed by each effective scan point in the reference expansion region 502 is determined to be the shape of the rod.

[0137] In the above implementation method, by determining the shape of each expansion point in the reference expansion region, the spatial distribution characteristics of the expansion points can be understood, which helps to identify the real objects in the reference expansion region.

[0138] S53: If a real object is detected, the region where the real object is located is removed from the reference dilation region to obtain the effective dilation region in the second direction around the high-reflection region.

[0139] If a real object is detected, it means that the expansion points in the reference expansion region are not all caused by the high-inflection expansion phenomenon. In order to avoid the real object's location being mistakenly marked during subsequent labeling and thus providing incorrect point cloud results to the upper-layer perception algorithm, the region where the real object is located is removed from the reference expansion region to obtain the effective expansion region in the second direction around the high-inflection region, so that there is no region corresponding to the real object in the effective expansion region in the second direction around the high-inflection region.

[0140] For example, when the effective expansion region around the high-reflection region includes an effective expansion region in a second direction around the high-reflection region, the implementation of step 203 above includes the following steps S61 to S63:

[0141] S61: Determine the upper limit of the second expansion rate of the high-reflectivity region based on the global reflectivity characteristic value.

[0142] There is a pre-defined second correspondence between the global reflectivity feature value and the second dilation rate upper limit value. By querying this second correspondence, the second dilation rate upper limit value corresponding to the global reflectivity feature value can be obtained. The second dilation rate upper limit value represents the theoretical maximum dilation rate of the high reflectivity region when the effective dilation region in the second direction in the current frame is normal, under the global reflectivity feature value condition of the previous frame.

[0143] In some embodiments, a target interval containing the global reflectivity feature value of the previous frame can be determined from multiple preset intervals. A second correspondence is then queried based on the target interval to obtain the upper limit of the second dilation rate corresponding to the target interval. The second correspondence includes the relationship between the preset interval and the upper limit of the second dilation rate; the larger the upper limit of the preset interval, the smaller the corresponding upper limit of the first dilation rate.

[0144] For example, the global reflectivity feature value is specifically the global reflectivity mean (avg_ref_ratio), which is the reflectivity mean calculated based on the reflectivity measurements of the remaining scan points. The second expansion rate upper limit applies to the entire high-reflectivity region and can therefore also be called the high-reflectivity region expansion rate upper limit. Based on this, the second correspondence can be shown in Table 2 below:

[0145] Table 2

[0146] Global mean reflectance Upper limit of expansion rate in high-reflection regions 0<avg_ref_ratio<=8 0.5 8<avg_ref_ratio<=14 0.45 14<avg_ref_ratio<=20 0.4 20<avg_ref_ratio<26 0.35 26<avg_ref_ratio<255 0.3

[0147] Referring to Table 2, if the target interval of the mean global reflectance of the previous frame is (0, 8], then the upper limit of the expansion rate of the high reflectance region is 0.5.

[0148] S62: Calculate the expansion rate of the high-reflection region based on the number of high-reflection points within the high-reflection region and the number of expansion points within the effective expansion region in the second direction surrounding the high-reflection region.

[0149] After determining the high-reflection region and the effective expansion region in the second direction surrounding the high-reflection region, the number of high-reflection points in the high-reflection region (denoted as the first number) and the number of expansion points in the effective expansion region in the second direction surrounding the high-reflection region (denoted as the second number) are counted. The ratio obtained by dividing the second number by the first number is taken as the expansion rate of the high-reflection region.

[0150] S63: If the expansion rate of the high-reflection region is greater than the upper limit of the second expansion rate, then the effective expansion region in the second direction around the high-reflection region is determined to be abnormal.

[0151] After calculating the expansion rate of the high-reflection region, the expansion rate is compared with the upper limit of the second expansion rate. If the expansion rate of the high-reflection region is greater than the upper limit of the second expansion rate, the effective expansion region in the second direction around the high-reflection region is determined to be abnormal. Conversely, if the expansion rate of the high-reflection region is less than or equal to the upper limit of the second expansion rate, the effective expansion region in the second direction around the high-reflection region is determined to be normal.

[0152] For example, assuming the average global reflectance of the previous frame is 23, then according to Table 2 above, the upper limit of the expansion rate of the high-reflectance region corresponding to 23 is 0.35. If the calculated expansion rate of the high-reflectance region is 0.5, since 0.5 is greater than 0.35, it is determined that the effective expansion region in the second direction surrounding the high-reflectance region is abnormal.

[0153] For example, when the effective expansion region around the high-reflection region includes an effective expansion region in a second direction around the high-reflection region, the above step 203 is implemented as follows: S71 to S73:

[0154] S71: Determine the threshold for the number of expansion points in high-reflectivity regions based on the global reflectivity characteristic value.

[0155] There is a pre-defined third correspondence between the global reflectivity feature value and the inflation point number threshold. By querying this third correspondence, the inflation point number threshold corresponding to the global reflectivity feature value can be obtained. The inflation point number threshold is the upper limit of the number of inflation points, representing the maximum theoretical number of inflation points that can exist in the normal effective inflation region in the second direction in the current frame under the global reflectivity feature value of the previous frame.

[0156] In some embodiments, a target interval containing the global reflectivity feature value of the previous frame can be determined from multiple preset intervals. A third correspondence is then queried based on the target interval to obtain the corresponding inflation point threshold. The third correspondence includes the relationship between the preset intervals and the inflation point threshold; the larger the upper limit of the preset interval, the larger the corresponding inflation point threshold.

[0157] For example, the global reflectance feature value is specifically the global reflectance mean (avg_ref_ratio). Based on this, the third correspondence can be shown in Table 3 below:

[0158] Table 3

[0159] Global mean reflectance Inflation point number threshold 0<avg_ref_ratio<=8 100 8<avg_ref_ratio<=14 120 14<avg_ref_ratio<=20 150 20<avg_ref_ratio<26 600 26<avg_ref_ratio<255 10000

[0160] Referring to Table 3, if the target interval of the mean global reflectance of the previous frame is (0, 8], then the threshold for the number of expansion points is 100.

[0161] S72: Determine the number of expansion points within the effective expansion region in the second direction surrounding the high-reflection region.

[0162] After determining the effective expansion region in the second direction around the high-reflection region, count the number of expansion points in the effective expansion region.

[0163] S73: If the number of expansion points in the effective expansion region in the second direction is greater than the expansion point number threshold, then the effective expansion region in the second direction around the high-reflection region is determined to be abnormal.

[0164] For example, assuming the average global reflectance of the previous frame is 23, then according to Table 3 above, the threshold for the number of dilatation points corresponding to 23 is 600. If the number of dilatation points in the effective dilatation area is 650, since 650 is greater than 600, it is determined that the effective dilatation area in the second direction around the high reflectance area is abnormal.

[0165] In this implementation, considering that even if an extremely large high-reflectivity area (such as an extremely large high-reflectivity sign) expands by only a few rows, the visual expansion phenomenon is quite severe, but the expansion rate of the high-reflectivity area is relatively low, in this case, the effective expansion area in the second direction surrounding the high-reflectivity area might be determined to be normal because the expansion rate of the high-reflectivity area is lower than the upper limit of the second expansion rate. Therefore, a judgment condition is adopted that the number of expansion points in the effective expansion area in the second direction is greater than the number of expansion points threshold. This ensures that even if the expansion rate of the high-reflectivity area is lower than the upper limit of the second expansion rate, the effective expansion area in the second direction surrounding the high-reflectivity area might still be considered abnormal because the number of expansion points is greater than the number of expansion points threshold, thus improving the accuracy and comprehensiveness of identifying whether the effective expansion area in the second direction is abnormal.

[0166] It should be noted that, in the above embodiments, two methods are provided for determining whether the effective expansion region in the second direction surrounding the high-reflection region is abnormal: one based on the expansion rate of the high-reflection region and the other based on the number of expansion points. In specific implementation, if the result of either method is that the effective expansion region in the second direction surrounding the high-reflection region is abnormal, then it can be determined that the effective expansion region in the second direction surrounding the high-reflection region is abnormal.

[0167] In step 204, the anomalies in the effective expansion region include: anomalies in the effective expansion region along a first direction surrounding the high-reflection region, and / or anomalies in the effective expansion region along a second direction surrounding the high-reflection region. In specific implementations, there may be cases where only the effective expansion region along the first direction is abnormal, cases where only the effective expansion region along the second direction is abnormal, or cases where both the effective expansion regions along the first and second directions are abnormal. Regardless of which direction the effective expansion region is abnormal, the high-reflection region and the abnormal effective expansion region can be marked using a preset marking method, such as using a rectangular bounding box, to facilitate identification by the upper-layer perception algorithm. This rectangular bounding box can be the smallest bounding rectangle of the abnormal effective expansion region.

[0168] Figure 6 This is a schematic diagram of two-dimensional marking of high-reflectivity regions and abnormal effective expansion regions provided in an embodiment of this application.

[0169] For example, such as Figure 6 As shown, the high-reflectivity region 601 and the corresponding effective expansion region 602 are marked in two dimensions. For example, the minimum bounding rectangle of the effective expansion region 602 is used as the marked region 603. The coordinates of the four vertices of the marked region 603 are as follows: top left (pitch1, yaw1, dist_a), top right (pitch2, yaw2, dist_a), bottom left (pitch4, yaw4, dist_b), and bottom right (pitch3, yaw3, dist_b). The warning flag marked within the high-reflectivity region 601 can be understood as a warning sign, such as a high-reflectivity sign.

[0170] like Figure 6 As shown, the current frame point cloud of the lidar consists of t rows and n columns. Assuming the preset boundary limit is the m-th row, then... Figure 6The first m rows are the statistical region, and rows m+1 to t are the non-statistical region. The distance measurement value of the high-reflection point on the upper boundary of the high-reflection region 601 is dist_a, and the distance measurement value of the high-reflection point on the lower boundary of the high-reflection region 601 is dist_b. The upper boundary of the effective expansion region 602 is top row, the lower boundary is down row, the left boundary is left row, and the right boundary is right row.

[0171] For example, Figure 6 The marking method shown is two-dimensional marking; in specific implementations, three-dimensional marking can also be used. For example, [the following text is incomplete and requires further context to translate accurately]. Figure 6 The marked region 603 in the image is mapped to a three-dimensional region in space. The four vertices of the marked region 603 may not be on the same plane when mapped to the three-dimensional region. Therefore, the four vertices of the marked region 603 can be mapped to the three-dimensional region and converted into eight vertices to form a three-dimensional bounding box.

[0172] For example, the method of converting the four vertices of the marked region into eight vertices after mapping to the three-dimensional region can include the following steps S81 to S83:

[0173] S81: Convert the coordinates of the four vertices of the marked area from polar coordinates to Cartesian coordinates, and obtain the coordinates of the four vertices in Cartesian coordinates as A (xa, ya, za), B (xb, yb, zb), C (xc, yc, zc), and D (xd, yd, zd).

[0174] S82: Given the coordinates of four points A, B, C, and D, determine the minimum and maximum values ​​in the X, Y, and Z directions, respectively: x_min, x_max, y_min, y_max, z_min, and z_max.

[0175] Where x_min and x_max are the minimum and maximum x-coordinates of points A, B, C, and D, respectively. y_min and y_max are the minimum and maximum y-coordinates of points A, B, C, and D, respectively. z_min and z_max are the minimum and maximum z-coordinates of points A, B, C, and D, respectively.

[0176] S83: Based on the preset margin and the above x_min, x_max, y_min, y_max, z_min, z_max, obtain the coordinates of the eight vertices after transformation.

[0177] If the preset margin is represented as pad, then the coordinates of the eight vertices are represented as follows:

[0178] a(x_min-pad, y_min-pad, z_min-pad);

[0179] b(x_min-pad, y_max+pad, z_min-pad);

[0180] c(x_min-pad, y_max+pad, z_max+pad);

[0181] d(x_min-pad, y_min-pad, z_max+pad);

[0182] e(x_max-pad, y_min-pad, z_min-pad);

[0183] f(x_max+pad, y_max+pad, z_min-pad);

[0184] g(x_max+pad, y_max+pad, z_max+pad);

[0185] h(x_max+pad, y_min-pad, z_max+pad);

[0186] The coordinates of the eight vertices constitute a three-dimensional bounding box, which is used to mark the high-reflection region and the effective expansion region of the anomaly.

[0187] Figure 7 This is a high-reflection dilatation point cloud map for rainy weather provided in an embodiment of this application. (By...) Figure 7 It can be seen that each of the high-reflectivity cards 701 to 704 is surrounded by a layer of point cloud with low reflectivity, and the point cloud with low reflectivity is an expansion point.

[0188] Figure 8 This is a labeled, rainy, high-reflection dilatation top-down point cloud map provided in an embodiment of this application. Combined with... Figure 7 and Figure 8 , Figure 8 The four 3D bounding boxes from left to right are: the top-view 3D bounding box used to mark the high-reflection sign 701 and the effective expansion area around it; the top-view 3D bounding box used to mark the high-reflection sign 702 and the effective expansion area around it; the top-view 3D bounding box used to mark the high-reflection sign 703 and the effective expansion area around it; and the top-view 3D bounding box used to mark the high-reflection sign 704 and the effective expansion area around it.

[0189] Figure 9 This is a marked, rainy, high-inverse-expansion frontal point cloud map provided in an embodiment of this application. Combined with... Figure 7 and Figure 9 , Figure 9The four 3D bounding boxes from left to right are: the frontal 3D bounding box used to mark the high-reflection sign 701 and the effective expansion area around it; the frontal 3D bounding box used to mark the high-reflection sign 702 and the effective expansion area around it; the frontal 3D bounding box used to mark the high-reflection sign 703 and the effective expansion area around it; and the frontal 3D bounding box used to mark the high-reflection sign 704 and the effective expansion area around it.

[0190] Figure 10 This is a flowchart illustrating another point cloud processing method for lidar provided in an embodiment of this application.

[0191] For example, such as Figure 10 As shown, the point cloud processing method for LiDAR includes steps 1001 to 1010, which are part of the point cloud processing system within the LiDAR data processing system. The method uses distance and reflectivity measurements from the point cloud data uploaded by the LiDAR for processing. The following explanation uses an example where the LiDAR scanning method is line-by-line scanning and the global reflectivity feature is the global reflectivity average:

[0192] Step 1001: Input the point cloud data of the current frame.

[0193] Step 1002: Scan line by line to obtain the high-inversion lines and the corresponding expansion points.

[0194] Specifically, the point cloud data of the current frame is scanned line by line to obtain each high inverse line in the current frame. The region formed by the dilated points corresponding to the high inverse lines can be understood as the dilated sub-regions on the left and right sides of the high inverse lines mentioned above.

[0195] Step 1003: Determine whether the expansion point corresponding to the high reflectivity line is abnormal based on the global reflectivity mean of the previous frame.

[0196] Here, the high-reflection row can be understood as the high-reflection sub-region mentioned above, and this step can be implemented based on S41 to S42 above. Determining whether the expansion point corresponding to the high-reflection row is abnormal can be understood as: determining whether the expansion sub-region corresponding to the high-reflection row is normal.

[0197] Step 1004: Determine whether there is a rod-shaped object below the high-reflection area.

[0198] Step 1005: Determine if there is a wall below the high-reflection area.

[0199] Among them, see Figure 5 The method shown is used to determine whether a rod-shaped object exists below the high-reflection area. See also... Figure 4 The method shown determines whether there is a wall below the high-reflection area.

[0200] Step 1006: Track the complete high-reflection region and the effective expansion region around the high-reflection region.

[0201] The complete high-reflectivity region is obtained by merging high-reflectivity lines whose field-of-view overlap is greater than a preset overlap threshold. The effective expansion region surrounding the high-reflectivity region includes the effective expansion regions in the first and second directions surrounding the high-reflectivity region.

[0202] Step 1007: Calculate the expansion rate of the high-reflection region.

[0203] Step 1008: Determine whether the expansion of the high-reflectivity region is abnormal based on the global reflectivity mean of the previous frame.

[0204] Specifically, determining whether the expansion of the high-reflection region is abnormal can be understood as: determining whether the effective expansion region below the high-reflection region is abnormal. The determination method can be found in S61 to S63 and S71 to S72 above. For the effective expansion region that is determined to be abnormal, the high-reflection region and the corresponding effective expansion region are marked. The marking method can be the three-dimensional bounding box marking method mentioned above, which will not be repeated here.

[0205] Step 1009: Calculate the mean global reflectance of the current frame.

[0206] Step 1010: Store the average global reflectance of the current frame as the information input for the next frame.

[0207] In this embodiment, by introducing the global reflectivity feature value of the previous frame, a reliable reference standard is provided for determining whether abnormal dilation exists in the current frame. When an anomaly is determined in the effective dilation region, marking the high reflectivity region and the effective dilation region can effectively provide an early warning of performance degradation of the lidar, allowing the upper-layer perception algorithm to combine these markings and avoid misidentification. By identifying whether there are rod-shaped objects and walls below the high reflectivity region and removing rod-shaped objects and walls in the effective dilation region, it avoids marking real objects that should not be marked during subsequent marking, providing accurate reference data for the upper-layer perception algorithm. The marked point cloud data provides clear guidance for subsequent processing, helping the control algorithm to make more accurate decisions.

[0208] Figure 11 This is a schematic diagram of the structure of a lidar provided in an embodiment of this application.

[0209] For example, such as Figure 11 As shown, the lidar 1100 includes a memory 1101 and a processor 1102. The memory 1101 stores executable program code 11011, and the processor 1102 is used to call and execute the executable program code 11011 to perform a point cloud processing method for lidar.

[0210] Furthermore, this application also protects an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to perform a point cloud processing method for lidar provided in this application.

[0211] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0212] It should be understood that the apparatus provided in this embodiment is used to perform the above-described point cloud processing method for lidar, and therefore can achieve the same effect as the above-described implementation method.

[0213] When using an integrated unit, the device may include a processing module and a storage module. Specifically, when the device is applied to a LiDAR, the processing module can be used to control and manage the LiDAR's operations. The storage module can be used to support the LiDAR in executing relevant program code.

[0214] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits shown in conjunction with the disclosure of this application. The processor may also be a combination of functions that implement computing capabilities, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0215] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a point cloud processing method for a lidar provided in the above embodiments.

[0216] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the point cloud processing method for lidar provided in the above embodiment.

[0217] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to implement the point cloud processing method for lidar provided in the above embodiment.

[0218] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0219] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0220] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0221] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A point cloud processing method for lidar, characterized in that, The method includes: Acquire the point cloud data of the current frame collected by the lidar and the global reflectivity feature value of the previous frame of the current frame; Based on the point cloud data of the current frame, determine the high-reflectivity region in the current frame and the effective dilatation region around the high-reflectivity region; Based on the global reflectivity characteristic value, determine whether the effective expansion region around the high reflectivity region is abnormal; If the effective expansion region is determined to be abnormal, then the high-reflection region and the effective expansion region are marked.

2. The method according to claim 1, characterized in that, The step of determining the high-reflectivity region in the current frame based on the point cloud data of the current frame includes: Based on the point cloud data of the current frame, determine whether there is a high-reflectivity sub-region in each scanned sub-region, and obtain each high-reflectivity sub-region existing in the current frame; wherein, the high-reflectivity sub-region includes n consecutive high-reflectivity points, and the high-reflectivity points are scan points that meet the preset high-reflectivity point requirements, where n is greater than the preset number; The high-reflectivity sub-regions with a field-of-view overlap greater than a preset overlap threshold are merged to obtain the high-reflectivity region in the current frame.

3. The method according to claim 2, characterized in that, The effective expansion region around the high-reflectivity region includes: an effective expansion region around the high-reflectivity region in a first direction, where the first direction is the scanning direction of the lidar; Determining the effective dilatation region around the high-reflectivity region based on the point cloud data of the current frame includes: For each of the high-reflectivity sub-regions present in the current frame, based on the point cloud data of the current frame, it is determined whether there is an expanded sub-region in the first direction surrounding the high-reflectivity sub-region, thus obtaining the expanded sub-regions corresponding to each of the high-reflectivity sub-regions in the current frame; wherein, the scan points in the expanded sub-regions are all expanded points that meet the preset expansion point requirements; Based on the dilated sub-regions corresponding to each of the high-reflectivity sub-regions in the current frame, the effective dilated region in the first direction surrounding the high-reflectivity region is determined.

4. The method according to claim 3, characterized in that, The step of determining whether the effective expansion region surrounding the high-reflectivity region is abnormal based on the global reflectivity feature value includes: Based on the global reflectivity characteristic value, determine the upper limit of the first expansion rate of the high reflectivity sub-region; For each of the high-reflectivity sub-regions, the expansion rate of the high-reflectivity sub-region is calculated. If the expansion rate of the high-reflectivity sub-region is greater than the first expansion rate upper limit, then the expansion sub-region corresponding to the high-reflectivity sub-region is determined to be abnormal.

5. The method according to claim 4, characterized in that, The calculation of the expansion rate of the high-inflection subregion includes: Determine the number of high-reflection points in the high-reflection sub-region and the number of expansion points in the corresponding expansion sub-region; The ratio obtained by dividing the number of expansion points by the number of high-reflection points is used as the expansion rate of the high-reflection sub-region.

6. The method according to claim 1, characterized in that, The effective expansion region around the high-reflectivity region includes: an effective expansion region in a second direction around the high-reflectivity region, wherein the second direction is perpendicular to the scanning direction of the lidar; Determining the effective dilatation region around the high-reflectivity region based on the point cloud data of the current frame includes: Based on the point cloud data of the current frame, a reference expansion region is determined in a second direction around the high-reflectivity region; wherein the boundary of the reference expansion region in the second direction is the same as the boundary of the high-reflectivity region, and the boundary of the reference expansion region in the scanning direction is determined based on the location of an expansion point that meets a preset expansion point requirement. Identify the presence of a real object within the reference expansion region; If a real object is detected, the area where the real object is located is removed from the reference expansion region to obtain an effective expansion region in the second direction surrounding the high-reflectivity region.

7. The method according to claim 6, characterized in that, The step of identifying whether a real object exists within the reference expansion region includes: Determine the number of theoretical scan points and the number of effective scan points within the reference expansion region; Calculate the ratio of the number of effective scan points to the number of theoretical scan points; If the ratio is greater than a preset ratio threshold, then it is determined that a real object exists within the reference expansion region.

8. The method according to claim 6, characterized in that, The step of identifying whether a real object exists within the reference expansion region includes: Determine the shape formed by the expansion points within the reference expansion region; Determine whether the shape is a preset shape of a real object; If the shape is the preset shape, then it is determined that a real object exists within the reference expansion region.

9. The method according to claim 6, characterized in that, The step of determining whether the effective expansion region surrounding the high-reflectivity region is abnormal based on the global reflectivity feature value includes: Based on the global reflectivity characteristic value, determine the upper limit of the second expansion rate of the high-reflectivity region; The expansion rate of the high-reflection region is calculated based on the number of high-reflection points within the high-reflection region and the number of expansion points within the effective expansion region in the second direction surrounding the high-reflection region. If the expansion rate of the high-reflectivity region is greater than the upper limit of the second expansion rate, then the effective expansion region in the second direction surrounding the high-reflectivity region is determined to be abnormal.

10. The method according to claim 6, characterized in that, The step of determining whether the effective expansion region surrounding the high-reflectivity region is abnormal based on the global reflectivity feature value includes: Based on the global reflectivity characteristic value, determine the threshold for the number of expansion points in the high-reflectivity region; Determine the number of expansion points within the effective expansion region in the second direction surrounding the high-reflection region; If the number of expansion points within the effective expansion region in the second direction is greater than the expansion point number threshold, then the effective expansion region in the second direction surrounding the high-reflection region is determined to be abnormal.

11. The method according to any one of claims 1 to 10, characterized in that, If the current frame is not the first frame, the global reflectance feature value of the previous frame is calculated in the following way: The remaining scan points are obtained by removing scan points whose reflectance measurement value is greater than the first reflectance threshold and scan points whose reflectance measurement value is 0 from the point cloud data of the previous frame. Based on the reflectance measurements of the remaining scan points, the average reflectance of the remaining scan points is calculated, and the average reflectance is used as the global reflectance feature value of the previous frame.

12. A lidar, characterized in that, The lidar includes: Memory, used to store executable program code; A processor is configured to call and run the executable program code from the memory, causing the lidar to perform the method as described in any one of claims 1 to 11.

13. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the method as described in any one of claims 1 to 11.

Images