Passable area denoising method and device, computer equipment and storage medium
By identifying and removing noise points in a global coordinate system, the problem of LiDAR misidentifying obstacles under rain, snow, or dust conditions is solved, thus improving the environmental perception and safety of autonomous vehicles.
Patent Information
- Application Number
- CN202511546525.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-27
- Publication Date
- 2026-03-06
AI Technical Summary
In autonomous driving, lidar is prone to misidentifying small objects as obstacles in rainy, snowy, or dusty conditions, causing vehicles to decelerate or stop unexpectedly, reducing operational efficiency and passenger comfort. How can we effectively reduce noise in passable areas?
By acquiring the initial point cloud data of the vehicle, performing coordinate transformation, and then identifying and deleting noise points in the global coordinate system, noise points are filtered out using methods such as position distance, speed information, and echo intensity to ensure the accuracy of identification and the effectiveness of noise reduction.
It improves the quality and reliability of passable area data, helps vehicles perceive the environment more accurately, avoids unnecessary operations caused by noise interference, and enhances the safety and efficiency of autonomous driving.
Smart Images
Figure CN121616481A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of autonomous driving technology, and specifically to a method, apparatus, computer equipment, and storage medium for noise reduction in passable areas. Background Technology
[0002] In autonomous driving, passable areas can be regarded as a kind of general obstacle detection, effectively supplementing the boundary detection of road boundaries, unknown obstacles and irregular obstacles, providing important perception input for the path planning and obstacle avoidance of autonomous vehicles, and providing a safety net for vehicle driving.
[0003] Passable areas can be extracted from point cloud data acquired by LiDAR, millimeter-wave radar, or ultrasonic radar, or from image data acquired by cameras. Traditional rule-based algorithms and emerging deep learning are both effective means of obtaining passable areas. Among them, mass-produced vehicle-mounted LiDAR is mostly based on time-of-flight ranging, which has good ranging accuracy and has been widely used. However, LiDAR also encounters many challenges in practical use. Lasers have a small divergence angle and good directivity, meaning that the laser beam can still maintain a relatively concentrated and small spot after traveling a long distance. Therefore, small objects that are not of concern to the vehicle may be identified as obstacles, causing the boundary of the passable area to shrink. For example, in rainy or snowy weather conditions, obvious laser reflection points may be generated on raindrops and snowflakes; in dusty road sections or weather conditions, large particles in the air will also produce obvious reflections. This can cause unintended deceleration or even braking of autonomous vehicles, reducing the operating efficiency of autonomous vehicles.
[0004] Therefore, how to perform noise reduction on passable areas has become an urgent problem to be solved. Summary of the Invention
[0005] In view of this, the present invention provides a method, apparatus, computer device and storage medium for denoising passable areas, in order to solve the problem of how to denoise passable areas.
[0006] In a first aspect, the present invention provides a method for denoising passable areas, the method comprising: Obtain the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system; Perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system; Identify the backup point cloud data and determine the point cloud data to be identified from the backup point cloud data; The point cloud data to be identified is identified to determine whether the point cloud data to be identified contains noise. If the point cloud data to be identified is noisy, then the point cloud data to be identified will be deleted from the passable area of the current frame.
[0007] The denoising method for passable areas provided in this application embodiment obtains initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system. The initial point cloud data is transformed to obtain backup point cloud data in the global coordinate system, allowing for processing of the backup point cloud data in the global coordinate system and ensuring the accuracy of the processing. Then, the backup point cloud data is identified to determine the point cloud data to be identified, thus filtering the backup point cloud data and ensuring the accuracy of the determined point cloud data to be identified, thereby reducing the amount of data processing. The point cloud data to be identified is then identified to determine whether it contains noise. Based on the identification result, noise can be determined from the point cloud data to be identified. If the point cloud data to be identified is noise, it is deleted from the passable area of the current frame, thereby purifying the data, improving the quality and reliability of the passable area data, and enabling the vehicle to more accurately perceive the surrounding environment and make safer and more reasonable driving decisions, such as avoiding unnecessary braking or steering operations caused by noise interference. It achieves noise reduction processing for passable areas.
[0008] In one optional implementation, identifying the backup point cloud data, and determining the point cloud data to be identified from the backup point cloud data, includes: Obtain the location information of the backup point cloud data; Calculate the positional distance between the location information of the backup point cloud data and the location information of fixed obstacles in the map; If the distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, the backup point cloud data will be identified as the point cloud data to be identified.
[0009] The denoising method for passable areas provided in this application obtains the location information of backup point cloud data; calculates the location distance between the location information of the backup point cloud data and the location information of fixed obstacles in the map, ensuring the accuracy of the calculated location distance between the backup point cloud data and the fixed obstacles. If the location distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, the backup point cloud data is determined as the point cloud data to be identified, ensuring the accuracy of determining the backup point cloud data as the point cloud data to be identified, thereby avoiding the identification of the backup point cloud data corresponding to the fixed obstacle as noise, which would affect the safe driving of the vehicle.
[0010] In one optional implementation, the point cloud data to be identified is identified to determine whether it contains noise, including: Based on the location information of each point cloud data to be identified in the passable area of the current frame, the current point cloud data area to be identified is generated. Divide the current point cloud data region to be identified into multiple current frame grids; For each current frame grid, detect whether there is point cloud data to be identified in the current frame grid; If there is point cloud data to be identified in the current frame grid, then based on the position information of the current frame grid in the global coordinate system, the historical frame grid corresponding to the current frame grid is determined from at least one historical point cloud data region corresponding to the current point cloud data region. Detect whether historical point cloud data exists in each historical frame grid; Based on the detection results, determine whether the point cloud data to be identified is noise.
[0011] The denoising method for passable areas provided in this application generates a current point cloud data region based on the position information of each point cloud data to be identified within the passable area of the current frame. This ensures that the generated current point cloud data region contains only the point cloud data to be identified, thereby reducing the amount of data. Furthermore, it separates the point cloud data to be identified from other point cloud data, thus avoiding identification errors. The current point cloud data region to be identified is divided into multiple current frame grids. For each current frame grid, the presence of point cloud data to be identified is detected to ensure the accuracy of the detection results. If point cloud data to be identified exists in the current frame grid, the historical frame grid corresponding to the current frame grid is determined from at least one historical point cloud data region corresponding to the current point cloud data region, based on the position information of the current frame grid in the global coordinate system. This ensures that the determined historical frame grid and the current frame grid are at the same position in the global coordinate system, and thus, the detection of whether the point cloud data to be identified in the current frame grid is noise can be performed based on the current frame grid and the historical frame grid. Then, it checks whether historical point cloud data exists in each historical frame grid; based on the detection results, it determines whether the point cloud data to be identified is noise, thus ensuring the accuracy of the determined detection results.
[0012] In one optional implementation, the current point cloud data region to be identified is divided into multiple current frame grids, including: Obtain the speed information of the vehicle itself; Based on the speed information, determine the first division size; Identify the point cloud data to be identified in the current point cloud data area to be identified, and determine the sparse and dense areas of the point cloud in the current point cloud data area to be identified. Based on the first partitioning size, the sparse region of the point cloud is divided into multiple current frame grids; Based on the second partitioning size, the dense region of the point cloud is divided into multiple current frame grids; the second partitioning size is smaller than the first partitioning size.
[0013] The denoising method for passable areas provided in this application embodiment obtains the speed information of the vehicle. Based on the speed information, a first segmentation size is determined, thereby ensuring the accuracy of the determined first segmentation size and avoiding inaccurate recognition results due to an excessively large first segmentation size, or excessive data processing volume due to an excessively small first segmentation size. Then, the point cloud data to be identified in the current point cloud data area is identified to determine the sparse and dense regions of the point cloud, ensuring the accuracy of the determined sparse and dense regions. Based on the first segmentation size, the sparse region of the point cloud is divided into multiple current frame grids, ensuring the accuracy of the segmentation of the sparse region. Based on a second segmentation size, the dense region of the point cloud is divided into multiple current frame grids, ensuring the accuracy of the segmentation of the dense region, and thus ensuring the accuracy of the identification of the dense region.
[0014] In one optional implementation, determining whether the point cloud data to be identified is noise based on the detection results includes: Based on the detection results, determine the frequency of occurrence of the point cloud data to be identified; Based on the frequency of occurrence, determine whether the point cloud data to be identified is noise.
[0015] The denoising method for passable areas provided in this application determines the occurrence frequency of the point cloud data to be identified based on the detection results, ensuring the accuracy of the determined occurrence frequency. Based on the occurrence frequency, it determines whether the point cloud data to be identified is noise, ensuring the accuracy of the determination of whether the point cloud data to be identified is noise.
[0016] In one optional implementation, determining whether the point cloud data to be identified is noise based on its frequency of occurrence includes: Obtain the first echo intensity information corresponding to the point cloud data to be identified in the current frame grid and the second echo intensity information corresponding to the historical point cloud data in each historical frame grid; Calculate the average echo intensity information between the first echo intensity information and each second echo intensity information; Based on the average echo intensity information, determine the weight information corresponding to the occurrence frequency; The target frequency is obtained by multiplying the frequency of occurrence by the corresponding weight information. If the target frequency is less than the preset frequency threshold, the point cloud data to be identified is determined to be noise.
[0017] The décoration method for passable areas provided in this application obtains the first echo intensity information corresponding to the point cloud data to be identified in the current frame grid and the second echo intensity information corresponding to the historical point cloud data in each historical frame grid. The average echo intensity information between the first echo intensity information and each second echo intensity information is calculated, ensuring that the calculated average echo intensity information can characterize the echo intensity of the point cloud data to be identified in the current frame grid and the historical point cloud data in each historical frame grid. This avoids errors caused by considering only the point cloud data to be identified in the current frame grid or only the second echo intensity information corresponding to the historical point cloud data in each historical frame grid. Then, based on the average echo intensity information, the weight information corresponding to the occurrence frequency is determined, ensuring the accuracy of the determined weight information corresponding to the occurrence frequency. The occurrence frequency is multiplied by the corresponding weight information to obtain the target frequency, ensuring the accuracy of the determined target frequency. If the target frequency is less than a preset frequency threshold, the point cloud data to be identified is determined to be noise, ensuring the accuracy of the result of determining the point cloud data to be identified as noise.
[0018] In one optional implementation, obtaining the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system includes: Acquire the current raw point cloud data of the vehicle in its own coordinate system; Identify the current raw point cloud data and determine the boundary point cloud data corresponding to the current passable area; Based on the boundary point cloud data, determine the initial point cloud data of the passable area in the current frame in the vehicle coordinate system.
[0019] The denoising method for passable areas provided in this application acquires the current raw point cloud data of the vehicle in its own coordinate system; identifies the current raw point cloud data to determine the boundary point cloud data corresponding to the current passable area, ensuring the accuracy of the determined boundary point cloud data. Based on the boundary point cloud data, the initial point cloud data of the passable area in the current frame in the vehicle coordinate system is determined, ensuring the accuracy of the determined initial point cloud data.
[0020] In a second aspect, the present invention provides a noise reduction device for passable areas, the device comprising: The acquisition module is used to acquire the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system; The transformation module is used to perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system. The first identification module is used to identify the backup point cloud data and determine the point cloud data to be identified from the backup point cloud data. The second recognition module is used to recognize the point cloud data to be recognized and determine whether the point cloud data to be recognized is noise. The deletion module is used to delete the point cloud data to be identified from the passable area of the current frame if the point cloud data to be identified is noisy.
[0021] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the détente denoising method of the first aspect or any corresponding embodiment thereof.
[0022] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the décor denoising method for the first aspect or any corresponding embodiment thereof.
[0023] Fifthly, the present invention provides a computer program product, including computer instructions for causing a computer to execute the décor denoising method for the first aspect or any corresponding embodiment thereof. Attached Figure Description
[0024] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0025] Figure 1 This is a flowchart illustrating a method for denoising passable areas according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating another method for denoising passable areas according to an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the matching of backup point cloud data with fixed obstacles according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating another method for denoising passable areas according to an embodiment of the present invention; Figure 5 (i) is a schematic diagram of the original passable area obtained by lidar during driving in snowy weather according to an embodiment of the present invention; Figure 5 (ii) is a schematic diagram of the result of "denoising" processing without using map information according to an embodiment of the present invention; Figure 5 (iii) is a schematic diagram of a passable area obtained by using guardrail information in a high-precision map according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a noise reduction device for passable areas according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0027] In autonomous driving, passable areas can be regarded as a kind of general obstacle detection, effectively supplementing the boundary detection of road boundaries, unknown obstacles and irregular obstacles, providing important perception input for the path planning and obstacle avoidance of autonomous vehicles, and providing a safety net for vehicle driving.
[0028] Passable areas can be extracted from point cloud data acquired by LiDAR, millimeter-wave radar, or ultrasonic radar, or from image data acquired by cameras. Traditional rule-based algorithms and emerging deep learning are both effective means of obtaining passable areas. Among them, mass-produced vehicle-mounted LiDAR is mostly based on time-of-flight ranging, which has good ranging accuracy and has been widely used. However, LiDAR also encounters many challenges in practical use. Lasers have a small divergence angle and good directivity, meaning that the laser beam can still maintain a relatively concentrated and small spot after traveling a long distance. Therefore, small objects that are not of concern to the vehicle may be identified as obstacles, causing the boundary of the passable area to shrink. For example, in rainy or snowy weather conditions, obvious laser reflection points may be generated on raindrops and snowflakes; in dusty road sections or weather conditions, large particles in the air will also produce obvious reflections. This can cause unintended deceleration or even braking of autonomous vehicles, reducing the operating efficiency of autonomous vehicles and passenger comfort.
[0029] Therefore, how to perform noise reduction on passable areas has become an urgent problem to be solved.
[0030] It should be noted that the method for denoising traversable areas provided in this application embodiment can be executed by a device for denoising traversable areas. This device can be implemented as part or all of a computer device through software, hardware, or a combination of both. The computer device can be a server or a terminal. In this application embodiment, the server can be a single server or a server cluster composed of multiple servers. The terminal in this application embodiment can be a smartphone, personal computer, tablet computer, wearable device, or other intelligent hardware device such as an intelligent robot. The following method embodiments will use a computer device as the execution subject for explanation.
[0031] According to an embodiment of the present invention, a method for denoising traversable areas is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.
[0032] This embodiment provides a method for denoising passable areas, which can be used in the aforementioned computer equipment. Figure 1 This is a flowchart of a method for denoising passable areas according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system.
[0033] Specifically, the computer device can receive raw point cloud data from a LiDAR installed at at least one location on the vehicle. After obtaining the raw point cloud data, the computer device can perceive the entire environment corresponding to the raw point cloud data to distinguish between passable and impassable areas. The computer device can achieve this using various methods, such as semantic segmentation algorithms based on deep learning. The raw point cloud data is input into a trained neural network model, which can output the category of each point, such as distinguishing between ground (belonging to passable areas), buildings, vehicles, pedestrians, etc. (often belonging to impassable or avoidable areas). Alternatively, it can be based on traditional geometric feature analysis, such as classifying points within a certain height range (usually corresponding to the ground height range) as points related to passable areas based on the height distribution of the point cloud, while points above a certain height may be obstacles.
[0034] After determining the passable area, the computer device filters out the relevant data of those points belonging to the passable area from the overall collected raw point cloud data of the current frame, thereby obtaining the initial point cloud data of the passable area in the vehicle coordinate system of the current frame.
[0035] Step S102: Perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system.
[0036] Specifically, the computer device can acquire the vehicle's heading angle and position information in the global coordinate system. Then, based on the vehicle's heading angle information in the global coordinate system, a rotation matrix is generated, and based on the vehicle's position information in the global coordinate system, a translation vector is generated.
[0037] Then, the computer equipment uses the initial point cloud data to multiply by the rotation matrix and add the translation vector to obtain the backup point cloud data in the global coordinate system.
[0038] For example, when the vehicle is operating in a high-precision positioning state, its heading HAg and position (xg, yg) in the global coordinate system are obtained. The initial point cloud data of the passable area in the vehicle's coordinate system is Ω. S ={P s1 P s2 P s3 P s4 ...}, project the passable area in the vehicle coordinate system onto the global coordinate system to obtain the backup point cloud data Ω. G ={P g1 P g2 P g3 P g4 Specifically, the initial point cloud data of the passable area in the vehicle coordinate system can be projected to the global coordinate system using the following coordinate transformation:
[0039] in, , indicating the th in the traversable region in the global coordinate system. i One backup point cloud data, This indicates the first traversable area in the vehicle coordinate system. i Initial point cloud data, R It is a rotation matrix. T It is a translation vector. Rotation matrix. R It can be represented as:
[0040] Translation vector T It can be represented as:
[0041] Step S103: Identify the backup point cloud data and determine the point cloud data to be identified from the backup point cloud data.
[0042] Specifically, the computer equipment can identify the backup point cloud data and determine its location information. Then, based on the location information of the backup point cloud data, the point cloud data to be identified is determined within the backup point cloud data.
[0043] This step will be explained in detail below, and will not be repeated here.
[0044] Step S104: Identify the point cloud data to be identified and determine whether the point cloud data to be identified is noise.
[0045] Specifically, the computer device can identify the point cloud data to be identified, and determine the location information of the point cloud data in the passable area of the current frame and the historical passable area. Then, based on the location information of the point cloud data to be identified in the passable area of the current frame and the historical passable area, it is determined whether the point cloud data to be identified is noise.
[0046] This step will be explained in detail below, and will not be repeated here.
[0047] Step S105: If the point cloud data to be identified is noise, then delete the point cloud data to be identified from the passable area of the current frame.
[0048] Specifically, if the point cloud data to be identified is noisy, the computer device will delete the point cloud data to be identified from the passable area of the current frame in order to perform noise reduction processing on the passable area of the current frame.
[0049] The denoising method for passable areas provided in this application embodiment obtains initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system. The initial point cloud data is transformed to obtain backup point cloud data in the global coordinate system, allowing for processing of the backup point cloud data in the global coordinate system and ensuring the accuracy of the processing. Then, the backup point cloud data is identified to determine the point cloud data to be identified, thus filtering the backup point cloud data and ensuring the accuracy of the determined point cloud data to be identified, thereby reducing the amount of data processing. The point cloud data to be identified is then identified to determine whether it contains noise. Based on the identification result, noise can be determined from the point cloud data to be identified. If the point cloud data to be identified is noise, it is deleted from the passable area of the current frame, thereby purifying the data, improving the quality and reliability of the passable area data, and enabling the vehicle to more accurately perceive the surrounding environment and make safer and more reasonable driving decisions, such as avoiding unnecessary braking or steering operations caused by noise interference. It achieves noise reduction processing for passable areas.
[0050] This embodiment provides a method for denoising passable areas, which can be used in the aforementioned computer equipment. Figure 2 This is a flowchart of a method for denoising passable areas according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system.
[0051] Specifically, step S201 above may include the following steps: Step S2011: Obtain the current raw point cloud data of the vehicle in the vehicle coordinate system.
[0052] Specifically, the computer equipment can receive data from sensors such as lidar installed at at least one location on the vehicle, which collect data at a predetermined scanning frequency and angle range to obtain the current raw point cloud data.
[0053] Step S2012: Identify the current raw point cloud data and determine the boundary point cloud data corresponding to the current passable area.
[0054] Specifically, the computer device can identify the current raw point cloud data, and then perform cluster analysis based on the geometric features of the current raw point cloud data (such as distance, angle, density, etc.) to separate the point clouds belonging to passable areas and obstacles, and then extract the boundary point cloud data of the passable areas.
[0055] Step S2013: Based on the boundary point cloud data, determine the initial point cloud data of the passable area in the current frame in the vehicle coordinate system.
[0056] Specifically, the computer device determines the boundary point cloud data and the point cloud data inside it as the initial point cloud data of the passable area in the current frame in the vehicle coordinate system based on the boundary point cloud data.
[0057] Step S202: Perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system.
[0058] Please refer to the above description of step S201 for details on this step, which will not be repeated here.
[0059] Step S203: Identify the backup point cloud data and determine the point cloud data to be identified from the backup point cloud data.
[0060] Specifically, step S203 above may include the following steps: Step S2031: Obtain the location information of the backup point cloud data.
[0061] Specifically, the computer equipment identifies the backup point cloud data and determines its position information in the global coordinate system.
[0062] Step S2032: Calculate the positional distance between the positional information of the backup point cloud data and the positional information of the fixed obstacles in the map.
[0063] Specifically, computer equipment obtains the location information of fixed obstacles in the global coordinate system based on map data.
[0064] Then, the computer equipment calculates the positional distance between the location information of the backup point cloud data and the location information of the fixed obstacles in the map based on a preset distance calculation algorithm.
[0065] The preset distance calculation algorithm can be Euclidean distance, Manhattan distance, or Chebyshev distance. This application does not specifically limit the preset distance calculation algorithm.
[0066] Step S2033: If the positional distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, then the backup point cloud data is determined as the point cloud data to be identified.
[0067] Specifically, the computer device can compare the positional distance between the backup point cloud data and the fixed obstacle with a preset distance threshold. If the positional distance between the backup point cloud data and the fixed obstacle is greater than the preset distance threshold, the backup point cloud data is identified as the point cloud data to be identified, that is, the point cloud data to be identified may be noise.
[0068] If the distance between the backup point cloud data and the fixed obstacle is less than or equal to a preset distance threshold, then the backup point cloud data is determined to be the point cloud data corresponding to the fixed obstacle.
[0069] For example, a computer device can match fixed obstacles in a high-precision map using distance. Characterizing backup point cloud data and known fixed obstacles The distance, if point With the nearest fixed obstacle If the distance is less than the threshold d, then the boundary point is retained. Among them, distance It can be Euclidean distance. For example... Figure 3 The diagram illustrates the matching of backup point cloud data with fixed obstacles. If the distance between a solid point and fixed obstacle A or B is less than a threshold, the match is successful, and the boundary point is retained. If the distance between a hollow point and fixed obstacle A or B is greater than the threshold, the match fails, and the point cloud data is identified as the point cloud data to be identified, thus entering the denoising process.
[0070] Step S204: Identify the point cloud data to be identified and determine whether the point cloud data to be identified is noise.
[0071] Please refer to the above description of step S104 for details on this step, which will not be repeated here.
[0072] Step S205: If the point cloud data to be identified is noise, then the point cloud data to be identified is deleted from the passable area of the current frame.
[0073] Please refer to the above description of step S105 for details on this step, which will not be repeated here.
[0074] The denoising method for passable areas provided in this application acquires the current raw point cloud data of the vehicle in its own coordinate system; identifies the current raw point cloud data to determine the boundary point cloud data corresponding to the current passable area, ensuring the accuracy of the determined boundary point cloud data. Based on the boundary point cloud data, the initial point cloud data of the passable area in the current frame in the vehicle coordinate system is determined, ensuring the accuracy of the determined initial point cloud data.
[0075] Furthermore, the location information of backup point cloud data is acquired; the location distance between the backup point cloud data and the location information of fixed obstacles on the map is calculated, ensuring the accuracy of the calculated location distance between the backup point cloud data and the fixed obstacles. If the location distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, the backup point cloud data is identified as the point cloud data to be identified, ensuring the accuracy of identifying the backup point cloud data as the point cloud data to be identified. This avoids identifying the backup point cloud data corresponding to the fixed obstacle as noise, which would affect the safe driving of the vehicle.
[0076] This embodiment provides a method for denoising passable areas, which can be used in the aforementioned computer equipment. Figure 4 This is a flowchart of a method for denoising passable areas according to an embodiment of the present invention, such as... Figure 4 As shown, the process includes the following steps: Step S301: Obtain the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system.
[0077] Please refer to the above description of step S201 for details on this step, which will not be repeated here.
[0078] Step S302: Perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system.
[0079] Please refer to the above description of step S202 for details on this step, which will not be repeated here.
[0080] Step S303: Identify the backup point cloud data and determine the point cloud data to be identified from the backup point cloud data.
[0081] Please refer to the above description of step S203 for details on this step, which will not be repeated here.
[0082] Step S304: Identify the point cloud data to be identified and determine whether the point cloud data to be identified is noise.
[0083] Specifically, step S304 above may include the following steps: Step S3041: Generate the current point cloud data region to be identified based on the location information of each point cloud data to be identified in the passable area of the current frame.
[0084] Specifically, the computer device can filter the point cloud data to be identified based on the location information of each point cloud data in the passable area of the current frame, retaining only the point cloud data to be identified. Thus, based on the point cloud data to be identified, a current point cloud data region is generated.
[0085] Step S3042: Divide the current point cloud data region to be identified into multiple current frame grids.
[0086] Specifically, step S3042 above may include the following steps: Step a1: Obtain the speed information of the vehicle.
[0087] Specifically, computer equipment can obtain the vehicle's speed information based on wheel speed sensors.
[0088] Step a2: Determine the first division size based on the speed information.
[0089] Specifically, the computer device can determine the first division size corresponding to the speed information based on the correspondence between the speed information and the division size.
[0090] Step a3: Identify the point cloud data to be identified in the current point cloud data area to be identified, and determine the sparse and dense point cloud areas in the current point cloud data area to be identified.
[0091] Specifically, the computer device can identify the point cloud data to be identified in the current point cloud data area, and perform clustering operations on the point cloud data to be identified based on a distance clustering algorithm to determine the sparse and dense areas of the point cloud in the current point cloud data area.
[0092] Step a4: Based on the first partitioning size, divide the sparse region of the point cloud into multiple current frame grids.
[0093] Specifically, the computer device can divide the sparse region of the point cloud into multiple current frame grids.
[0094] Step a5: Based on the second partitioning size, divide the dense region of the point cloud into multiple current frame grids.
[0095] Specifically, the second division size is smaller than the first division size. The computer device can determine the second division size based on the first division size. Optionally, the second division size can be half of the first division size, or the second division size can be one-third of the first division size. This application embodiment does not specifically limit the second division size.
[0096] Then, the computer device can divide the dense region of the point cloud into multiple current frame grids based on the second partitioning size.
[0097] Step S3043: For each current frame grid, detect whether there is point cloud data to be identified in the current frame grid.
[0098] Specifically, for each current frame grid, the computer device can determine the presence of point cloud data to be identified by counting the number of points cloud data to be identified in each current frame grid. If the number of points cloud data to be identified in the current frame grid is greater than 0, it is determined that there are points cloud data to be identified in the current frame grid. If the number of points cloud data to be identified in the current frame grid is 0, there are no points cloud data to be identified in the current frame grid, and no further operation is required.
[0099] Step S3044: If there is point cloud data to be identified in the current frame grid, then based on the position information of the current frame grid in the global coordinate system, determine the historical frame grid corresponding to the current frame grid from at least one historical point cloud data region corresponding to the current point cloud data region.
[0100] Specifically, if point cloud data to be identified exists in the current frame grid, the computer device acquires the position information of the current frame grid in the global coordinate system. Then, based on the position information of the current frame grid in the global coordinate system, the historical frame grid corresponding to the current frame grid is determined in at least one historical frame of point cloud data to be identified, corresponding to the current point cloud data region.
[0101] The current frame grid has the same position information as each historical frame grid in global coordinates.
[0102] Step S3045: Detect whether historical point cloud data exists in each historical frame grid.
[0103] Specifically, the computer equipment sequentially counts whether historical point cloud data exists in each historical frame grid.
[0104] Step S3046: Based on the detection results, determine whether the point cloud data to be identified is noise.
[0105] Specifically, step S3046 above may include the following steps: Step b1: Based on the detection results, determine the occurrence frequency of the point cloud data to be identified.
[0106] Specifically, if historical point cloud data exists in the historical frame grid, the count is 1; if historical point cloud data does not exist in the historical frame grid, the count is 0. The computer device counts the proportion of 0 in the current frame grid and each historical frame grid. Then, by subtracting the proportion of 0 in the current frame grid and each historical frame grid from 1, the frequency of occurrence of the point cloud data to be identified is obtained.
[0107] For example, suppose the computer device counts the historical frame grids in the previous 5 frames corresponding to the current frame grid. Suppose that 3 of these historical frame grids contain historical point cloud data, then they are counted as 1 respectively; the remaining 2 historical frame grids do not contain historical point cloud data, then they are counted as 0 respectively. Since 0 accounts for one-third of the total in the current frame grid and all historical frame grids, the occurrence frequency of the point cloud data to be identified is two-thirds.
[0108] Step b2: Determine whether the point cloud data to be identified is noise based on the frequency of occurrence.
[0109] In one optional implementation, the computer device can compare the occurrence frequency of the point cloud data to be identified with a preset frequency threshold. If the occurrence frequency is less than the preset frequency threshold, the point cloud data to be identified is determined to be noise. If the occurrence frequency is greater than or equal to the preset frequency threshold, the point cloud data to be identified is determined not to be noise.
[0110] For example, in the global coordinate system, a square grid region Grid of length L with side length L is divided around the vehicle as the center. Any cell within this region can be represented as... Draw the boundary points of the drivable area in the grid region. The frequency plot, statistically analyzed within a defined grid. The frequency of occurrence of inner boundary points. Assuming the frequency of the model output is f, within a continuous time t, the mesh... If an inner boundary point appears k times, then its frequency is... It can be represented as:
[0111] For fixed infrastructure, its location is fixed in the global coordinate system. Therefore, within a defined location range, the probability of its corresponding passable area boundary point appearing is theoretically close to 1. However, for random noise, the probability of its appearance is much lower than 1. Setting an appropriate threshold... ,when When the probability of filtering out is lower than The passable area boundary points are removed, i.e., passable area boundary points that appear randomly in consecutive frames.
[0112] In another alternative implementation, step b2 above may include the following steps: Step b21: Obtain the first echo intensity information corresponding to the point cloud data to be identified in the current frame grid and the second echo intensity information corresponding to the historical point cloud data in each historical frame grid.
[0113] Specifically, the computer device can identify the echo intensity corresponding to the point cloud data to be identified in the current frame grid, and determine the first echo intensity information corresponding to the point cloud data to be identified in the current frame grid. Then, the computer device can identify the echo intensity corresponding to the historical point cloud data in each historical frame grid, and determine the second echo intensity information corresponding to the historical point cloud data in each historical frame grid.
[0114] Step b22: Calculate the average echo intensity information between the first echo intensity information and each second echo intensity information.
[0115] Specifically, the computer equipment can calculate the average echo intensity information between the first echo intensity information and each second echo intensity information.
[0116] Step b23: Determine the weight information corresponding to the occurrence frequency based on the average echo intensity information.
[0117] Specifically, computer equipment can determine the weight information corresponding to the average echo intensity information based on the correspondence between echo intensity and weight.
[0118] Step b24: Multiply the occurrence frequency by the corresponding weight information to obtain the target frequency.
[0119] Specifically, computer equipment can multiply the frequency of occurrence by the corresponding weight information to obtain the target frequency.
[0120] Step b25: If the target frequency is less than the preset frequency threshold, then the point cloud data to be identified is determined to be noise.
[0121] Specifically, the computer device can compare the target frequency with a preset frequency threshold. If the target frequency is less than the preset frequency threshold, the point cloud data to be identified is determined to be noise.
[0122] If the target frequency is greater than or equal to the preset frequency threshold, then the point cloud data to be identified is determined to be noisy.
[0123] Step S305: If the point cloud data to be identified is noise, then the point cloud data to be identified is deleted from the passable area of the current frame.
[0124] For details on this step, please refer to the description of step S205 above; it will not be repeated here.
[0125] For example, such as Figure 5 (i) represents the original passable area obtained by LiDAR during driving in snowy weather. EGO represents the autonomous vehicle. The black dots in the figure are the boundary points of the passable area, which are distributed at equal angles around the perimeter. The boundary points of the passable area in the two ellipses A and B are the "noise" introduced by the snowflakes, which causes the boundary of the passable area to shrink, resulting in the autonomous vehicle slowing down or stopping, thus reducing operational efficiency. Figure 5 (ii) is the result of direct "denoising" without using map information. The boundary points of the passable area within the rectangle become sparse. These are actually road guardrails, which are clear physical boundaries and should be preserved. Direct "denoising" would affect driving safety. Figure 5 (iii) shows the passable area obtained after using guardrail information from the high-precision map. It can be seen that the guardrail area is preserved, and the "snowflake" noise on the front and rear sides of the EGO is effectively suppressed, taking into account both driving safety and traffic efficiency.
[0126] The denoising method for passable areas provided in this application generates a current point cloud data region based on the position information of each point cloud data to be identified within the passable area of the current frame. This ensures that the generated current point cloud data region contains only the point cloud data to be identified, thereby reducing the amount of data. Furthermore, it separates the point cloud data to be identified from other point cloud data, thus avoiding recognition errors. Then, the speed information of the vehicle is acquired; based on the speed information, a first partitioning size is determined, ensuring the accuracy of the determined first partitioning size and avoiding inaccurate recognition results due to an excessively large first partitioning size, or excessive data processing volume due to an excessively small first partitioning size. Next, the point cloud data to be identified within the current point cloud data region is identified, determining sparse and dense point cloud regions within the current point cloud data region, ensuring the accuracy of the determined sparse and dense point cloud regions. Based on the first partitioning size, the sparse point cloud region is divided into multiple current frame grids, ensuring the accuracy of the sparse point cloud region division. Based on the second partitioning size, the dense region of the point cloud is divided into multiple current frame grids, which ensures the accuracy of partitioning the dense region of the point cloud, and thus ensures the accuracy of identifying the dense region of the point cloud.
[0127] For each current frame grid, the presence of point cloud data to be identified is checked to ensure the accuracy of the detection results. If point cloud data to be identified exists in the current frame grid, the corresponding historical frame grid is determined from at least one historical frame of point cloud data corresponding to the current point cloud data region, based on the position information of the current frame grid in the global coordinate system. This ensures that the determined historical frame grid and the current frame grid are at the same position in the global coordinate system, and thus, the point cloud data to be identified in the current frame grid can be detected as noise based on the current frame grid and the historical frame grid. Then, based on the detection results, the occurrence frequency of the point cloud data to be identified is determined, ensuring the accuracy of the determined occurrence frequency. The first echo intensity information corresponding to the point cloud data to be identified in the current frame grid and the second echo intensity information corresponding to the historical point cloud data in each historical frame grid are obtained. The average echo intensity information between the first echo intensity information and each second echo intensity information is calculated. This ensures that the calculated average echo intensity information can characterize the echo intensity of the point cloud data to be identified in the current frame grid and the historical point cloud data in each historical frame grid. This avoids errors caused by considering only the point cloud data to be identified in the current frame grid or only the second echo intensity information corresponding to the historical point cloud data in each historical frame grid. Then, based on the average echo intensity information, the weight information corresponding to the occurrence frequency is determined, ensuring the accuracy of the determined weight information corresponding to the occurrence frequency. The occurrence frequency is multiplied by the corresponding weight information to obtain the target frequency, ensuring the accuracy of the determined target frequency. If the target frequency is less than a preset frequency threshold, the point cloud data to be identified is determined to be noise, ensuring the accuracy of the result of determining the point cloud data to be identified to be noise.
[0128] This embodiment also provides a noise reduction device for passable areas, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, hardware implementations, or a combination of software and hardware, are also possible and contemplated.
[0129] This embodiment provides a noise reduction device for passable areas, such as Figure 6 As shown, it includes: The acquisition module 401 is used to acquire the initial point cloud data of the passable area corresponding to the vehicle in the current frame in the vehicle coordinate system; The conversion module 402 is used to perform coordinate transformation on the initial point cloud data to obtain backup point cloud data in the global coordinate system. The first identification module 403 is used to identify the backup point cloud data and determine the point cloud data to be identified in the backup point cloud data. The second recognition module 404 is used to recognize the point cloud data to be recognized and determine whether the point cloud data to be recognized is noise. The deletion module 405 is used to delete the point cloud data to be identified from the passable area of the current frame if the point cloud data to be identified is noisy.
[0130] In some optional implementations, the first identification module 403 is specifically used to acquire the location information of the backup point cloud data; calculate the location distance between the location information of the backup point cloud data and the location information of the fixed obstacle in the map; if the location distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, then the backup point cloud data is identified as the point cloud data to be identified.
[0131] In some optional implementations, the second identification module 404 is specifically used to generate a current point cloud data region to be identified based on the position information of each point cloud data to be identified in the passable area of the current frame; divide the current point cloud data region to be identified into multiple current frame grids; for each current frame grid, detect whether there is point cloud data to be identified in the current frame grid; if there is point cloud data to be identified in the current frame grid, determine the historical frame grid corresponding to the current frame grid from at least one historical point cloud data region corresponding to the current point cloud data region to be identified based on the position information of the current frame grid in the global coordinate system; detect whether there is historical point cloud data in each historical frame grid; and determine whether the point cloud data to be identified is noise based on the detection results.
[0132] In some optional implementations, the second identification module 404 is specifically used to acquire the speed information of the vehicle; determine a first division size based on the speed information; identify the point cloud data to be identified in the current point cloud data area to be identified, and determine the sparse and dense areas of the point cloud in the current point cloud data area to be identified; divide the sparse area of the point cloud into multiple current frame grids based on the first division size; divide the dense area of the point cloud into multiple current frame grids based on the second division size; the second division size is smaller than the first division size.
[0133] In some optional implementations, the second identification module 404 is specifically used to determine the occurrence frequency of the point cloud data to be identified based on the detection results; and to determine whether the point cloud data to be identified is noise based on the occurrence frequency.
[0134] In some optional implementations, the second identification module 404 is specifically used to acquire the first echo intensity information corresponding to the point cloud data to be identified in the current frame grid and the second echo intensity information corresponding to the historical point cloud data in each historical frame grid; calculate the average echo intensity information between the first echo intensity information and each second echo intensity information; determine the weight information corresponding to the occurrence frequency based on the average echo intensity information; multiply the occurrence frequency by the corresponding weight information to obtain the target frequency; if the target frequency is less than a preset frequency threshold, then the point cloud data to be identified is determined to be noise.
[0135] In some optional implementations, the acquisition module 401 is specifically used to acquire the current raw point cloud data collected by the vehicle in the vehicle coordinate system; identify the current raw point cloud data to determine the boundary point cloud data corresponding to the current passable area; and determine the initial point cloud data of the current frame passable area in the vehicle coordinate system based on the boundary point cloud data.
[0136] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.
[0137] In this embodiment, the denoising device for passable areas is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.
[0138] This invention also provides a computer device having the above-described features. Figure 6 The noise reduction device for the passable area is shown.
[0139] Please see Figure 7 , Figure 7 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 7As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 7 Take a processor 10 as an example.
[0140] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.
[0141] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.
[0142] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0143] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.
[0144] The computer device also includes an input device 30 and an output device 40. The processor 10, memory 20, input device 30, and output device 40 can be connected via a bus or other means. Figure 7 Taking the example of a connection between China and Israel via a bus.
[0145] Input device 30 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the computer device, such as a touchscreen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 40 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The aforementioned display devices include, but are not limited to, liquid crystal displays, light-emitting diodes, displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.
[0146] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.
[0147] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.
[0148] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.
Claims
1. A method of passable region denoising, the method comprising: The method comprises: obtaining initial point cloud data of a current frame passable area corresponding to a vehicle in a vehicle coordinate system; performing coordinate conversion on the initial point cloud data to obtain backup point cloud data in a global coordinate system; identifying the backup point cloud data to determine to-be-identified point cloud data in the backup point cloud data; identifying the to-be-identified point cloud data to determine whether the to-be-identified point cloud data is a noise point; if the to-be-identified point cloud data is a noise point, deleting the to-be-identified point cloud data from the current frame passable area.
2. The method of claim 1, wherein, The identifying the backup point cloud data to determine to-be-identified point cloud data in the backup point cloud data comprises: obtaining position information of the backup point cloud data; calculating a position distance between the position information of the backup point cloud data and position information of a fixed obstacle in a map; if the position distance between the backup point cloud data and the fixed obstacle is greater than a preset distance threshold, determining the backup point cloud data as the to-be-identified point cloud data.
3. The method of claim 1, wherein, The identifying the to-be-identified point cloud data to determine whether the to-be-identified point cloud data is a noise point comprises: generating a current to-be-identified point cloud data region based on position information of each to-be-identified point cloud data in the current frame passable area; dividing the current to-be-identified point cloud data region into a plurality of current frame grids; for each current frame grid, detecting whether the to-be-identified point cloud data exists in the current frame grid; if the to-be-identified point cloud data exists in the current frame grid, determining a historical frame grid corresponding to the current frame grid from at least one historical to-be-identified point cloud data region corresponding to the current to-be-identified point cloud data region according to position information of the current frame grid in a global coordinate system; detecting whether historical point cloud data exists in each historical frame grid; determining whether the to-be-identified point cloud data is a noise point according to the detection result.
4. The method of claim 3, wherein, The dividing the current to-be-identified point cloud data region into a plurality of current frame grids comprises: obtaining speed information of the vehicle; determining a first division size according to the speed information; identifying the to-be-identified point cloud data in the current to-be-identified point cloud data region to determine a point cloud sparse region and a point cloud dense region in the current to-be-identified point cloud data region; dividing the point cloud sparse region into a plurality of current frame grids based on the first division size; dividing the point cloud dense region into a plurality of current frame grids based on a second division size; the second division size is smaller than the first division size.
5. The method of claim 3, wherein, The determining whether the to-be-identified point cloud data is a noise point according to the detection result comprises: determining an occurrence frequency corresponding to the to-be-identified point cloud data according to the detection result; determining whether the to-be-identified point cloud data is a noise point according to the occurrence frequency.
6. The method of claim 5, wherein, The determining whether the to-be-identified point cloud data is a noise point according to the occurrence frequency comprises: obtaining first echo intensity information corresponding to to-be-identified point cloud data in the current frame grid and second echo intensity information corresponding to historical point cloud data in each historical frame grid; average echo intensity information between the first echo intensity information and each of the second echo intensity information; determining weight information corresponding to the occurrence frequency according to the average echo intensity information; multiplying the occurrence frequency by the corresponding weight information to obtain a target frequency; if the target frequency is less than a preset frequency threshold, determining that the to-be-identified point cloud data is a noise point.
7. The method of claim 1, wherein, The method comprises the following steps: acquiring initial point cloud data of a current frame passable area corresponding to a self-driving vehicle in a self-driving coordinate system; acquiring current original point cloud data collected by the self-driving vehicle in the self-driving coordinate system; identifying the current original point cloud data to determine boundary point cloud data corresponding to the current passable area; 8. A passable area denoising apparatus characterized by comprising: determining the initial point cloud data of the current frame passable area in the self-driving coordinate system according to the boundary point cloud data. The device comprises: an acquisition module configured to acquire initial point cloud data of a current frame passable area corresponding to a self-driving vehicle in a self-driving coordinate system; a conversion module configured to perform coordinate conversion on the initial point cloud data to obtain backup point cloud data in a global coordinate system; a first identification module configured to identify the backup point cloud data to determine to-be-identified point cloud data in the backup point cloud data; a second identification module configured to identify the to-be-identified point cloud data to determine whether the to-be-identified point cloud data is a noise point; 9. A computer device, comprising: a deletion module configured to delete the to-be-identified point cloud data from the current frame passable area if the to-be-identified point cloud data is a noise point. comprise:
10. A computer-readable storage medium, characterized in that, a memory and a processor, which are communicatively connected, and the memory stores computer instructions, and the processor executes the computer instructions to perform the passable area denoising method in any one of claims 1 to 7. The computer readable storage medium stores computer instructions, and the computer instructions are used to make a computer execute the passable area denoising method in any one of claims 1 to 7.