Interest region vehicle-mounted point cloud extraction method and device and vehicle

Abstract

The application discloses an interest area vehicle-mounted point cloud extraction method and device and a vehicle. The method comprises the following steps: firstly, determining a boundary polygon corresponding to a target vehicle based on a current position of the target vehicle and a preset map; secondly, extracting a target interest area of the target vehicle according to the boundary polygon and a preset interest range; thirdly, extracting a plurality of target grids located in the target interest area from a grid network of the target vehicle; and finally, selecting target vehicle-mounted point clouds from candidate vehicle-mounted point cloud data of the target vehicle by using the plurality of target grids. The application solves the technical problem that the selection accuracy of the interest area by the vehicle is low in the related art, which easily leads to the extraction of point cloud data of invalid interest areas by the vehicle, thereby increasing the calculation amount of the point cloud data extracted by the vehicle.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving, and more specifically, to a method, apparatus, and vehicle for extracting region of interest (ROI) onboard point clouds. Background Technology

[0002] In the field of autonomous driving, vehicles need to acquire point cloud data of scanned target objects within a very short time, such as a few hundred nanoseconds, using LiDAR. Then, they must rely on vision technology to perform obstacle segmentation, lane detection, and obstacle classification on the point cloud data. In other words, vehicles have high requirements for the real-time performance of point cloud data processing. Therefore, improving the accuracy of the vehicle's selection of regions of interest in the point cloud data to further reduce the computational load of point cloud data processing is particularly important.

[0003] In related technologies, point cloud data regions are typically divided into different grids, and point cloud data is filled into different grids. Then, by traversing the point cloud data in different grids, the point cloud data to be processed is further selected according to preset rules. Since point cloud data contains many points that do not need to be processed, the information redundancy is relatively large. Therefore, directly filling all point cloud data as regions of interest into the grid can easily lead to low accuracy in selecting regions of interest. This results in problems such as large computational load, low efficiency, and poor real-time performance in vehicle point cloud data processing.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a method, apparatus, and vehicle for extracting regions of interest (ROI) point clouds on a vehicle, thereby addressing the technical problem in related technologies where the vehicle's selection accuracy for ROI is low, easily leading to the extraction of point cloud data from invalid ROIs, thus increasing the computational load for point cloud data extraction.

[0006] According to one embodiment of the present invention, a method for extracting a region of interest (ROI) vehicle-mounted point cloud is provided, comprising: determining a boundary polygon corresponding to the target vehicle based on the current position of the target vehicle and a preset map, wherein the boundary polygon is used to describe the road range where the target vehicle is located; extracting a target ROI of the target vehicle according to the boundary polygon and a preset ROI, wherein the preset ROI is determined by the maximum scanning range of the target vehicle's solid-state radar; extracting multiple target grids located within the target ROI from the grid network of the target vehicle; and selecting a target vehicle-mounted point cloud from candidate vehicle-mounted point cloud data of the target vehicle using the multiple target grids, wherein the target vehicle-mounted point cloud includes at least one vehicle-mounted point cloud within the multiple target grids.

[0007] Optionally, determining the boundary polygon corresponding to the target vehicle based on the current location of the target vehicle and a preset map includes: obtaining the current location of the target vehicle; querying the map information of the target vehicle from the preset map based on the current location, wherein the preset map is a preset high-precision map; and extracting the boundary polygon corresponding to the target vehicle from the map information, wherein the boundary polygon is determined by multiple boundary plane points.

[0008] Optionally, extracting the target region of interest (ROI) of the target vehicle based on the boundary polygon and the preset ROI includes: calculating the convex hull range corresponding to the boundary polygon according to the preset convex hull algorithm; determining the preset ROI based on the maximum scanning range of the solid-state radar of the target vehicle; and determining the target ROI of the target vehicle based on the intersection area of ​​the convex hull range and the preset ROI.

[0009] Optionally, extracting multiple target graticles located within the target region of interest from the graticle network of the target vehicle includes: establishing a graticle network based on the target region of interest of the target vehicle, wherein the graticle network contains multiple candidate graticles; calculating the region coordinate span based on the target region of interest of the target vehicle to determine the main direction corresponding to the target region of interest; and extracting multiple target graticles located within the target region of interest from the multiple candidate graticles according to the main direction and the outer contour polygon of the target region of interest.

[0010] Optionally, establishing a grid network based on the target vehicle's target region of interest includes: initializing the bird's-eye view grid corresponding to the target vehicle according to the target vehicle's target region of interest, and establishing a grid network, wherein the grid network contains multiple candidate grids that are multiple candidate sector grids.

[0011] Optionally, extracting multiple target grids located within the target interest region from multiple candidate grids based on the main direction and the outer contour polygon of the target interest region includes: determining the outer contour polygon based on the outer contour point set of the target interest region; performing scanning calculations using a preset scanning method based on the contour polygon and the main direction to obtain the calculation results; and extracting multiple target sector grids located within the target interest region from multiple candidate sector grids according to the calculation results.

[0012] Optionally, selecting a target vehicle point cloud from candidate vehicle point cloud data using multiple target grids includes: acquiring candidate vehicle point cloud data of the target vehicle, wherein the candidate vehicle point cloud data includes multiple candidate vehicle point clouds; filling the multiple candidate vehicle point clouds into the target grid according to a preset clustering algorithm to obtain a filling result; and selecting the target vehicle point cloud from the multiple candidate vehicle point clouds based on the filling result.

[0013] Optionally, obtaining candidate vehicle-mounted point cloud data of the target vehicle includes: reading the original vehicle-mounted point cloud data frame from the solid-state radar of the target vehicle; converting the point cloud coordinates in the original vehicle-mounted point cloud data frame to the vehicle body coordinate system of the target vehicle; and performing point cloud filtering on the original vehicle-mounted point cloud data frame to obtain candidate vehicle-mounted point cloud data.

[0014] According to one embodiment of the present invention, an apparatus for extracting vehicle point clouds of a region of interest is also provided, comprising: a boundary determination module, configured to determine a boundary polygon corresponding to the target vehicle based on the current position of the target vehicle and a preset map, wherein the boundary polygon is used to describe the road range where the target vehicle is located; a region extraction module, configured to extract a target region of interest of the target vehicle according to the boundary polygon and a preset region of interest, wherein the preset region of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle; a grid extraction module, configured to extract multiple target grids located within the target region of interest from the grid network of the target vehicle; and a point cloud selection module, configured to select a target vehicle point cloud from candidate vehicle point cloud data of the target vehicle using multiple target grids, wherein the target vehicle point cloud includes at least one vehicle point cloud within the multiple target grids.

[0015] Optionally, the boundary determination module is also used to: obtain the current position of the target vehicle; based on the current position, query the map information of the target vehicle from a preset map, wherein the preset map is a preset high-precision map; extract the boundary polygon corresponding to the target vehicle from the map information, wherein the boundary polygon is determined by multiple boundary plane points.

[0016] Optionally, the region extraction module is also used to: calculate the convex hull range corresponding to the boundary polygon according to a preset convex hull algorithm; determine a preset range of interest based on the maximum scanning range of the solid-state radar of the target vehicle; and determine the target region of interest of the target vehicle according to the intersection area of ​​the convex hull range and the preset range of interest.

[0017] Optionally, the grid extraction module is also used to: establish a grid network based on the target region of interest of the target vehicle, wherein the grid network contains multiple candidate grids; calculate the region coordinate span based on the target region of interest of the target vehicle to determine the main direction corresponding to the target region of interest; and extract multiple target grids located within the target region of interest from multiple candidate grids according to the main direction and the outer contour polygon of the target region of interest.

[0018] Optionally, the grid extraction module is also used to: initialize the bird's-eye view grid corresponding to the target vehicle based on the target vehicle's target region of interest, and establish a grid network, wherein the grid network contains multiple candidate grids that are multiple candidate sector grids.

[0019] Optionally, the raster extraction module is also used to: determine the outer contour polygon based on the outer contour point set of the target region of interest; perform scanning calculations using a preset scanning method based on the contour polygon and the main direction to obtain the calculation results; and extract multiple target sector rasters located within the target region of interest from multiple candidate sector rasters according to the calculation results.

[0020] Optionally, the point cloud selection module is also used to: acquire candidate vehicle point cloud data of the target vehicle, wherein the candidate vehicle point cloud data includes multiple candidate vehicle point clouds; fill the multiple candidate vehicle point clouds into the target grid according to a preset clustering algorithm to obtain the filling result; and select the target vehicle point cloud from the multiple candidate vehicle point clouds based on the filling result.

[0021] Optionally, the point cloud selection module is also used to: read the original data frame of the vehicle-mounted point cloud from the solid-state radar of the target vehicle; transform the point cloud coordinates in the original data frame of the vehicle-mounted point cloud to the vehicle body coordinate system of the target vehicle; and perform point cloud filtering on the original data frame of the vehicle-mounted point cloud to obtain candidate vehicle-mounted point cloud data.

[0022] According to one embodiment of the present invention, a vehicle is also provided, including an on-board memory and an on-board processor. The on-board memory stores a computer program, and the on-board processor is configured to run the computer program. When the computer program is executed by the processor to extract the region of interest on-board point cloud, it can implement the steps in any of the above-described method embodiments.

[0023] In this embodiment of the invention, firstly, based on the current position of the target vehicle and a preset map, the boundary polygon corresponding to the target vehicle is determined. Then, based on the boundary polygon and a preset range of interest, the target region of interest (ROI) of the target vehicle is extracted. Next, multiple target grids located within the ROI are extracted from the grid network of the target vehicle. Finally, using the multiple target grids, the target vehicle point cloud is selected from the candidate vehicle point cloud data of the target vehicle. This achieves the goal of obtaining the target ROI by using the boundary polygon corresponding to the environment in which the target vehicle is located at its current position and the maximum scanning range of the solid-state radar. The target ROI is then used to simplify the point cloud data that the target vehicle needs to extract, thereby improving the accuracy of selecting the ROI of the point cloud data and reducing the computational workload of extracting the point cloud data. This solves the technical problem in related technologies where the accuracy of selecting the ROI of the vehicle is low, which easily leads to the vehicle extracting point cloud data from invalid ROIs, thereby increasing the computational workload of extracting the point cloud data. Attached Figure Description

[0024] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings:

[0025] Figure 1 This is a hardware structure block diagram of a vehicle terminal for implementing a method for extracting vehicle-mounted point clouds of regions of interest according to an embodiment of the present invention.

[0026] Figure 2 This is a flowchart of a method for extracting region of interest (ROI) vehicle-mounted point clouds according to an embodiment of the present invention;

[0027] Figure 3 This is a schematic diagram of the boundary polygon extracted from the intersection of an optional target vehicle according to an embodiment of the present invention;

[0028] Figure 4 This is a schematic diagram illustrating how an optional target vehicle determines a target region of interest based on a convex hull range and a preset region of interest, according to an embodiment of the present invention.

[0029] Figure 5 This is a schematic diagram of an optional initialization of a fan-shaped bird's-eye view grid according to an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram of an optional extraction of the outer contour of a target region of interest according to an embodiment of the present invention;

[0031] Figure 7 This is a schematic diagram of an optional extraction of multiple target grids according to an embodiment of the present invention;

[0032] Figure 8 This is a schematic diagram of an optional point cloud data filling method according to an embodiment of the present invention;

[0033] Figure 9 This is a flowchart of an optional method for extracting and filling point cloud data of a region of interest according to an embodiment of the present invention;

[0034] Figure 10 This is a structural block diagram of a vehicle-mounted point cloud extraction device for a region of interest according to one embodiment of the present invention. Detailed Implementation

[0035] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0036] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0037] According to an embodiment of the present invention, an embodiment of a method for extracting a region of interest (ROI) on-board point cloud 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.

[0038] The method embodiment provided in Embodiment 1 of the present invention can be executed in a power battery vehicle, a hybrid vehicle, or a similar computing device. Figure 1 This is a hardware structure block diagram of a vehicle terminal for implementing a method for extracting vehicle-mounted point clouds of regions of interest according to an embodiment of the present invention. Figure 1 As shown, the vehicle terminal 10 (or mobile device 10) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the vehicle terminal 1 described above. For example, the vehicle terminal 10 may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0039] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be implemented wholly or partially as software, hardware, firmware, or any other combination. Furthermore, the data processing circuits may be a single, independent processing module, or may be wholly or partially integrated into any other element within the vehicle terminal 10 (or mobile device). As involved in the embodiments of the present invention, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor terminal path connected to an interface).

[0040] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the method for extracting the region of interest (ROI) vehicle point cloud in this embodiment of the invention. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the above-mentioned method for extracting the region of interest (ROI) vehicle point cloud. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the vehicle terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0041] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the vehicle terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0042] Under the above operating environment, the present invention provides, as follows: Figure 2 The method for extracting vehicle-mounted point clouds of the region of interest is shown. Figure 2 This is a flowchart of a method for extracting region of interest (ROI) vehicle-mounted point clouds according to an embodiment of the present invention, as shown below. Figure 2 As shown, the method includes the following steps S1 to S4.

[0043] Step S1: Based on the current location of the target vehicle and the preset map, determine the boundary polygon corresponding to the target vehicle, wherein the boundary polygon is used to describe the road range where the target vehicle is located.

[0044] In step S1 above, the target vehicle is a vehicle traveling on the road. The current position is the target vehicle's current location information calculated and analyzed based on global positioning information received by a GPS receiver and inertial navigation data received by an inertial navigation system. The GPS receiver can be either the BeiDou Navigation Satellite System or the Global Positioning System (GPS). The preset map is a high-precision map. The boundary polygon is the geometric shape corresponding to the road, intersection, or tunnel entrance where the target vehicle is located, extracted from the preset map.

[0045] Based on the target vehicle's current location and a preset map representing a high-precision map, the system calculates and analyzes the boundary polygons corresponding to roads, intersections, or tunnel entrances in the environment where the target vehicle is currently located.

[0046] Step S2: Extract the target region of interest of the target vehicle based on the boundary polygon and the preset region of interest, wherein the preset region of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle.

[0047] In step S2 above, the preset area of ​​interest (OPI) is the maximum scanning range of the target vehicle centered on the solid-state radar. The scanning range is centered on the solid-state radar, with the scanning radius as the radius, and the horizontal direction as the starting point of the arc, with the arc angle representing the visible viewing angle. Whether the visible viewing angle calculated by the solid-state radar in the horizontal direction is clockwise or counterclockwise is set by the designer based on the actual situation. For illustrative purposes, the following use case is uniformly set to counterclockwise calculation, and the visible viewing angle of the solid-state radar is a full circle. The solid-state radar can be a solid-state lidar. The target region of interest (ROI) is the intersection between the convex hull range corresponding to the current position of the target vehicle and the preset OPI. The convex hull range is the boundary range of one or more boundary polygons corresponding to the environment where the target vehicle is currently located, calculated according to the preset convex hull algorithm. The preset convex hull algorithms can be Gift Wrapping, Graham Scan, Jarvis March, Quick Hull, Divide-and-Conquer, Monotone Chain, Kallay, and Marriage-before-Conquest.

[0048] The target vehicle first calculates and analyzes the convex hull range of one or more boundary polygons corresponding to information such as roads, intersections, and tunnel entrances in its current environment, as well as a preset region of interest determined by the maximum scanning range of the solid-state radar. The target vehicle then calculates the intersection between the convex hull range corresponding to its current position and the preset region of interest, and determines this intersection as the target region of interest.

[0049] Step S3: Extract multiple target grids located within the target region of interest from the grid network of the target vehicle.

[0050] In step S3 above, the raster network is a fan-shaped bird's-eye view raster created by first initializing a raster image based on the target region of interest for the target vehicle. The range of the raster network is typically the smallest circle that can cover the target region of interest. Each raster region in the raster network is a fan shape. Multiple target gratings are calculated and analyzed using the scanline method based on the raster network to determine the number of gratings occupied by the target region of interest on the raster network.

[0051] The target vehicle calculates and analyzes the multiple target grids occupied by the target region of interest on the grid network.

[0052] Step S4: Using multiple target grids, select a target vehicle point cloud from the candidate vehicle point cloud data of the target vehicle, wherein the target vehicle point cloud includes at least one vehicle point cloud within the multiple target grids.

[0053] In step S4 above, the candidate vehicle point cloud data is the point cloud data obtained by solid-state radar scanning the environment surrounding the target vehicle. The target vehicle point cloud is the point cloud data corresponding to the target region of interest selected by the target vehicle from the candidate vehicle point cloud data. The vehicle point cloud is a single candidate vehicle point cloud data.

[0054] The target vehicle reads a frame of point cloud data acquired by solid-state radar and uses this data as candidate vehicle-mounted point cloud data. This candidate point cloud data is then preprocessed. The preprocessing includes noise removal, removal of noisy points, and removal of invalid points. The target vehicle then converts the point cloud coordinates from the world coordinate system to the vehicle coordinate system, and then converts the X, Y, and Z axis coordinates to polar coordinates. Finally, the target vehicle uses a flood fill algorithm to fill the preprocessed target vehicle-mounted point cloud data into the aforementioned target grid.

[0055] Through the above steps S1 to S4, firstly, based on the current position of the target vehicle and the preset map, the boundary polygon corresponding to the target vehicle is determined. Then, based on the boundary polygon and the preset range of interest, the target region of interest of the target vehicle is extracted. Next, multiple target grids located within the target region of interest are extracted from the grid network of the target vehicle. Finally, using multiple target grids, the target vehicle point cloud is selected from the candidate vehicle point cloud data of the target vehicle. This achieves the goal of obtaining the target region of interest by using the boundary polygon corresponding to the environment in which the target vehicle is located at its current position and the maximum scanning range of the solid-state radar. Then, the target region of interest is used to simplify the point cloud data that the target vehicle needs to extract, thereby improving the accuracy of selecting the region of interest of the point cloud data and reducing the computational load of the vehicle to extract the point cloud data. This solves the technical problem in related technologies where the accuracy of the vehicle's selection of the region of interest is low, which easily leads to the vehicle extracting point cloud data from invalid regions of interest, thereby increasing the computational load of the vehicle to extract the point cloud data.

[0056] Optionally, in step S1, determining the boundary polygon corresponding to the target vehicle based on the current location of the target vehicle and the preset map further includes the following steps S11 to S13.

[0057] Step S11: Obtain the current location of the target vehicle.

[0058] In step S11 above, the target vehicle calculates and analyzes its current position based on the global positioning information and inertial navigation data obtained from the global positioning system receiver and the inertial navigation system.

[0059] Step S12: Based on the current location, retrieve the map information of the target vehicle from the preset map, where the preset map is a preset high-precision map.

[0060] In step S12 above, the target vehicle queries a preset map for high-precision map information corresponding to its current location. The high-precision map information retrieved by the target vehicle typically includes road information, intersection information, and tunnel entrance information where the target vehicle is currently located.

[0061] Step S13: Extract the boundary polygon corresponding to the target vehicle from the map information, wherein the boundary polygon is determined by multiple boundary plane points.

[0062] In step S13 above, the boundary plane points are simulated points representing objects in the high-precision map. The target vehicle calculates the corresponding boundary polygons in its current surrounding environment from the acquired high-precision map information. These boundary polygons represent the geometry of the roads, intersections, and tunnel entrances where the target vehicle is currently located. The boundary polygons are determined by multiple boundary plane points, whose coordinates are binary points, such as (x, y).

[0063] Figure 3 This is a schematic diagram of the boundary polygon extracted from the intersection of an optional target vehicle according to an embodiment of the present invention.

[0064] like Figure 3 As shown, the black dots are boundary plane points, representing objects surrounding the intersection where the target vehicle is currently located. The target vehicle is currently at an intersection. The target vehicle extracts high-precision map information corresponding to the current intersection from a preset map, then calculates the boundary polygon corresponding to the current intersection, and records this boundary polygon as the intersection polygon. The specific geometry of the intersection polygon is determined by the boundary plane points in the diagram.

[0065] When the target vehicle is in a complex terrain, such as a road and a tunnel entrance, the boundary polygon corresponding to the target vehicle's current position includes multiple boundary polygons, such as the boundary polygon of the corresponding road and the boundary polygon of the corresponding tunnel entrance.

[0066] Optionally, in step S2, extracting the target region of interest for the target vehicle based on the boundary polygon and the preset region of interest further includes the following steps S21 to S23.

[0067] Step S21: Calculate the convex hull range corresponding to the boundary polygon according to the preset convex hull algorithm.

[0068] In step S21 above, the preset convex hull algorithm is a convex hull algorithm selected by the designer based on the actual situation. The target vehicle uses the preset convex hull algorithm to calculate the convex hull range of one or more boundary polygons corresponding to its current position. These boundary polygons might include, for example, the intersection polygon and the road polygon corresponding to the target vehicle's current position.

[0069] Step S22: Determine the preset range of interest based on the maximum scanning range of the target vehicle's solid-state radar.

[0070] In step S22 above, the solid-state radar of the target vehicle is taken as the center, and the maximum scanning range of the solid-state radar is the preset range of interest.

[0071] Step S23: Determine the target region of interest for the target vehicle based on the intersection of the convex hull range and the preset region of interest.

[0072] In step S23 above, the target vehicle calculates the intersection region between the convex hull range obtained in step S21 and the preset interest range obtained in step S22, and uses the intersection region as the target interest region of the target vehicle.

[0073] Figure 4This is a schematic diagram illustrating how an optional target vehicle determines a target region of interest based on the convex hull range and a preset region of interest, according to an embodiment of the present invention.

[0074] like Figure 4 As shown, the target vehicle's current position is at an intersection, and there is only one boundary polygon corresponding to the current position, which is denoted as the intersection polygon. Figure 4 The polygons in the diagram represent intersection polygons, and the black dots are the boundary plane points that define these polygons. The circles represent preset areas of interest, denoted as the preset range, determined by the maximum scanning range of the target vehicle's solid-state radar. The preset convex hull algorithm uses a monotonic chain convex hull algorithm. The target vehicle's convex hull is calculated and analyzed using this algorithm, determining the convex hull range of the boundary polygon corresponding to its current position. Since the boundary polygon corresponding to the target vehicle's current position is only one intersection polygon, the calculated convex hull range for the target vehicle is a geometric shape consistent with the intersection polygon. Figure 4 The code also uses intersection polygons to represent the convex hull of the boundary polygon corresponding to the current position of the target vehicle. The target vehicle then calculates the intersection between the intersection polygon and the preset range; this calculated intersection is used as the target region of interest, such as... Figure 4 As shown in the shaded area.

[0075] Optionally, in step S3, extracting multiple target grids located within the target region of interest from the grid network of the target vehicle further includes the following steps S31 to S33.

[0076] Step S31: Based on the target interest region of the target vehicle, establish a grid network, wherein the grid network contains multiple candidate grids.

[0077] In step S31 above, the target vehicle establishes a minimum circular grid network that can cover the target region of interest. The grid network is a fan-shaped bird's-eye view grid. Each grid in the grid network is a fan, and each grid in the grid network is called a candidate grid, or a candidate fan-shaped grid.

[0078] Step S32: Based on the target vehicle's region of interest, calculate the region coordinate span and determine the main direction corresponding to the region of interest.

[0079] In step S32 above, the target vehicle calculates the principal direction of the target region of interest, which is the principal direction of the boundary polygon corresponding to the current position of the target vehicle, and also the principal direction of the boundary polygon obtained in step S1. The principal direction of the boundary polygon corresponding to the current position of the target vehicle is determined by the span between the boundary plane points in the X and Y axes, and is usually taken as the principal direction by the maximum span in the X and Y axes. The span of the boundary plane points in the X and Y axes is determined by subtracting the minimum value from the maximum value in the X and Y axes of these points.

[0080] Step S33: Based on the main direction and the outer contour polygon of the target interest region, extract multiple target rasters located within the target interest region from multiple candidate rasters.

[0081] In step S33 above, the target vehicle calculates and analyzes multiple target grids corresponding to the target region of interest in the grid network at the current position of the target vehicle based on the main direction and outer contour polygon of the target region of interest at the current position, through the preset scanning method in subsequent steps.

[0082] Optionally, in step S31, establishing a grid network based on the target vehicle's target region of interest further includes the following step S311.

[0083] Step S311: Based on the target vehicle's target region of interest, initialize the bird's-eye view grid corresponding to the target vehicle and establish a grid network, wherein the grid network contains multiple candidate grids that are multiple candidate sector grids.

[0084] In step S311 above, the target vehicle initializes the grid network by using the smallest circle as the area of ​​the grid network, based on the area of ​​interest that can be covered by the target vehicle. After the target vehicle initializes the grid network, a grid network composed of bird's-eye view grids is then established. The candidate sector grids are the candidate grids mentioned in step S31.

[0085] Figure 5 This is a schematic diagram of an optional initialization of a fan-shaped bird's-eye view grid according to an embodiment of the present invention.

[0086] like Figure 5 As shown, the target vehicle is currently located at an intersection. The polygon in the figure represents the boundary polygon of the intersection where the target vehicle is currently located, denoted as the intersection polygon. Using the solid-state radar as the center, the smallest circle covering the target's region of interest is used as the area of ​​the grid network. A grid network composed of bird's-eye view grids is initialized and established. Figure 5 The circular areas in the diagram represent the grid network established for the target vehicle. Each grid cell in the grid network is a sector-shaped area, for example, Figure 5 A sector-shaped grid area, as shown, represents a single grid cell. All grid cells in the grid network are sector-shaped, and each cell can be identified by its corresponding radius and angle. Whether the radius of the identifying grid cell is the radius of the upper or lower circle that makes up the grid is determined by the designer based on the actual situation; here, it is uniformly determined that the radius of the identifying grid cell is the upper circle radius. For example... Figure 5 As shown, a radius and an angle can identify a grid cell in a grid network, with the radius of the grid cell being the radius of the upper circle.

[0087] Optionally, in step S33, extracting multiple target gratings located within the target interest region from multiple candidate gratings based on the main direction and the outer contour polygon of the target interest region further includes the following steps S331 to S333.

[0088] Step S331: Determine the outer contour polygon based on the outer contour point set of the target region of interest.

[0089] In step S331 above, the target vehicle statistically analyzes the outer contour point set of the target interest region and calculates the outer contour polygon of the target interest region.

[0090] Figure 6 This is a schematic diagram of an optional method for extracting the outer contour of a target region of interest according to an embodiment of the present invention.

[0091] like Figure 6 As shown, the target vehicle is currently located at an intersection. The polygon in the figure is the boundary polygon corresponding to the intersection where the target vehicle is currently located, denoted as the intersection polygon. Figure 6 The circle in the diagram represents the preset area of ​​interest (OPI) for the target vehicle, the shaded area represents the target region of interest (PGI) for the target vehicle, and the black dots represent the boundary plane points that define the outer contour of the PPI. The target vehicle uses a preset convex hull algorithm to calculate the boundary polygon of the PPI, which is the boundary polygon of the shaded area in the diagram, and is denoted as the outer contour of the PPI.

[0092] Step S332: Based on the outline polygon and the main direction, a scanning calculation is performed using a preset scanning method to obtain the calculation result.

[0093] In step S332 above, the contour polygon is the outer contour polygon of the target region of interest determined in step S331. The preset scanning method is the surface line scanning method. Based on the outer contour polygon of the target region of interest and the main direction, the target vehicle calculates the grid cells occupied by the target region of interest in the grid network using the preset scanning method.

[0094] Step S333: Based on the calculation results, extract multiple target sector grids located within the target interest region from multiple candidate sector grids.

[0095] As mentioned in step S333 above, the multiple target sector grids refer to the multiple target grids mentioned in step S3. The target vehicle, based on the calculation results obtained in step S332, corresponds to multiple target grids in the grid network representing the target region of interest.

[0096] Figure 7 This is a schematic diagram of an optional extraction of multiple target gratings according to an embodiment of the present invention.

[0097] like Figure 7As shown, the target vehicle is currently located at an intersection. The polygon in the figure is the boundary polygon corresponding to the intersection where the target vehicle is currently located, denoted as the intersection polygon. Figure 7 The circular areas represent the grid network established for the target vehicle at its current position, where each sector is a grid cell. Black dots represent the boundary plane points of the polygon defining the target region of interest. The default scanning method uses the scanline method. The target vehicle uses the scanline method to calculate the number of grid cells occupied by its region of interest in the grid network. The shaded areas represent the number of grid cells occupied by the target region of interest in the grid network.

[0098] It is important to note that the target region of interest and the grid area occupied by the target region of interest in the raster network are not the same. For example, as Figure 7 As shown, there are also shaded areas outside the target region of interest because these shaded areas represent areas within the raster that are occupied by the target region of interest.

[0099] Optionally, in step S4, selecting the target vehicle point cloud from the candidate vehicle point cloud data using multiple target grids further includes the following steps S41 to S43.

[0100] Step S41: Obtain candidate vehicle point cloud data of the target vehicle, wherein the candidate vehicle point cloud data includes multiple candidate vehicle point clouds.

[0101] In step S41 above, the target vehicle acquires a frame of point cloud data from the solid-state radar as candidate vehicle-mounted point cloud data. Each individual point cloud data point is called a candidate vehicle-mounted point cloud. The candidate vehicle-mounted point cloud is the same as the vehicle-mounted point cloud in step S4 above.

[0102] Step S42: According to the preset clustering algorithm, fill multiple candidate vehicle point clouds into the target grid to obtain the filling result.

[0103] In step S42 above, the preset clustering algorithm is a clustering algorithm selected by the designer based on the actual situation, such as the flood filling clustering algorithm. The target vehicle uses the preset clustering algorithm to fill the candidate vehicle point cloud data corresponding to the target region of interest into the corresponding grid in the grid network.

[0104] Figure 8 This is a schematic diagram of an optional point cloud data filling method according to an embodiment of the present invention.

[0105] like Figure 8As shown, the circular area represents the grid network established by the target vehicle at its current position, and each sector in the figure is a grid cell within that network. The shaded area represents the point cloud data filled by the target vehicle, specifically the point cloud data corresponding to the target region of interest (ROI) from the point cloud data acquired by the target vehicle from solid-state radar. The grid cells filling the point cloud data are all grid cells that intersect with the target ROI; that is, the grid cells occupied by the target ROI within the grid network. The target vehicle fills the grid cells corresponding to its ROI with the point cloud data.

[0106] Step S43: Select the target vehicle point cloud from multiple candidate vehicle point clouds based on the filling results.

[0107] As in step S43 above, based on the filling result of step S42, the target vehicle extracts the point numbers of the candidate vehicle point clouds from the grid of multiple filled candidate vehicle point clouds and stores them. The candidate vehicle point clouds with extracted point numbers are used as the target vehicle point cloud. Since the point numbers of all filled candidate vehicle point clouds need to be extracted, and all filled candidate vehicle point clouds are point cloud data corresponding to the target region of interest, the target vehicle point cloud is also the point cloud data filled into the target region of interest. The point number of each candidate vehicle point cloud represents the information of that point cloud data, and the specific information extracted is set by the designer according to the actual situation.

[0108] The target vehicle extracts the point numbers of the candidate vehicle point cloud from the grid corresponding to the target region of interest in the grid network, and then stores these extracted point numbers for use by the target vehicle.

[0109] Optionally, in step S41, obtaining candidate vehicle point cloud data of the target vehicle further includes the following steps S411 to S413.

[0110] Step S411: Read the raw data frame of the on-board point cloud from the solid-state radar of the target vehicle.

[0111] In step S411 above, the vehicle point cloud raw data frame is a frame of vehicle point cloud raw data read by the target vehicle from the solid-state radar.

[0112] Step S412: Convert the point cloud coordinates in the original vehicle point cloud data frame to the vehicle body coordinate system of the target vehicle.

[0113] In step S412 above, the target vehicle transforms the coordinates of each point cloud data in the original vehicle point cloud data frame from the world coordinate system to the vehicle body coordinate system, and then to the polar coordinate system.

[0114] Step S413: Perform point cloud filtering on the original vehicle point cloud data frame to obtain candidate vehicle point cloud data.

[0115] In step S413 above, the target vehicle performs point cloud preprocessing operations such as point cloud filtering on each point cloud data in the original vehicle point cloud data frame to obtain candidate vehicle point cloud data.

[0116] The coordinate transformation of point cloud data in the original vehicle point cloud data frame by the target vehicle, as well as the filtering of point cloud data, both fall under the category of preprocessing of point cloud data by the target vehicle. Preprocessing of point cloud data also includes operations such as removing noise and invalid points from the point cloud data.

[0117] Figure 9 This is a flowchart of an optional method for extracting and filling point cloud data of a region of interest according to an embodiment of the present invention. Figure 9 As shown, the method includes the following steps:

[0118] Step P1: The target vehicle calculates its current location information using global positioning information and inertial navigation data;

[0119] Step P2: Based on the current location information, the target vehicle queries the high-precision map for the boundary polygons corresponding to the road, intersection, and tunnel entrance where the target vehicle is currently located.

[0120] Step P3: The target vehicle merges the boundary polygons corresponding to its current position;

[0121] Step P4: The target vehicle transforms the coordinates of the boundary polygon corresponding to its current position from the world coordinate system to the vehicle coordinate system.

[0122] Step P5: The target vehicle uses a monotonic chain convex hull algorithm (equivalent to the preset convex hull algorithm mentioned above) to calculate the convex hull range of the boundary polygon corresponding to the current position of the target vehicle. For example, if there are multiple boundary polygons corresponding to the current position of the target vehicle, then the target vehicle uses a monotonic chain convex hull algorithm to calculate the convex hull range of the multiple boundary polygons corresponding to the current position of the target vehicle.

[0123] Step P6: The target vehicle reads the scanning range of the solid-state radar (equivalent to the preset range of interest mentioned above). The scanning range of the solid-state radar is calculated counterclockwise with the solid-state radar as the center and the scanning radius as the radius, starting from the horizontal direction. This arc angle is used as the visible viewing angle of the solid-state radar. For example, if the scanning range of the solid-state radar is 100°, then the visible viewing angle of the solid-state radar is 360°.

[0124] Step P7: The target vehicle calculates the intersection between the solid-state radar scanning range and the convex hull range obtained in step P5, and uses this intersection as the region of interest (equivalent to the target region of interest mentioned above).

[0125] Step P8: The target vehicle uses the monotonic chain convex hull algorithm to calculate the boundary polygon of the region of interest.

[0126] Step P9: The target vehicle calculates the main direction of the merged boundary polygon obtained in step P3;

[0127] Step P10: The target vehicle uses the region of interest as a basis, takes the smallest circle that can cover the region of interest as the area of ​​the grid network, and initializes and establishes the grid network on the area of ​​the grid network.

[0128] Step P11: The target vehicle uses the scanline method (equivalent to the preset scanning method mentioned above) to calculate the number of grid cells occupied by the region of interest in the grid network.

[0129] Step P12: The target vehicle reads a frame of point cloud data obtained by the solid-state radar scanning the surrounding environment (equivalent to the above-mentioned vehicle-mounted point cloud raw data frame);

[0130] Step P13: The target vehicle performs preprocessing operations such as denoising, filtering, and invalid point removal on the point cloud data read in step P12.

[0131] Step P14: The target vehicle performs coordinate transformation on the point cloud data preprocessed in step P13. Specifically, it first transforms from the world coordinate system to the vehicle coordinate system, and then transforms from the vehicle coordinate system to the extreme value coordinate system.

[0132] In step P15, the target vehicle first uses the flood-fill clustering algorithm (equivalent to the aforementioned preset clustering algorithm) to fill the point cloud data corresponding to the region of interest into the grid cells occupied by the region of interest in the grid network after the preprocessing in step P13. The target vehicle then extracts the point sequence numbers of all the filled point cloud data and stores them for future use.

[0133] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0134] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0135] This embodiment also provides a device for extracting vehicle-mounted point clouds of regions of interest. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the terms "unit" and "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0136] Figure 10 This is a structural block diagram of a region-of-interest vehicle-mounted point cloud extraction device according to one embodiment of the present invention, as shown below. Figure 10 As shown, the vehicle-mounted point cloud extraction device 1000 for the region of interest includes:

[0137] The boundary determination module 1001 is used to determine the boundary polygon corresponding to the target vehicle based on the current position of the target vehicle and the preset map, wherein the boundary polygon is used to describe the road range where the target vehicle is located.

[0138] The region extraction module 1002 is used to extract the target region of interest of the target vehicle based on the boundary polygon and the preset region of interest, wherein the preset region of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle.

[0139] The grid extraction module 1003 extracts multiple target grids located within the target region of interest from the grid network of the target vehicle;

[0140] The point cloud selection module 1004 is used to select a target vehicle point cloud from candidate vehicle point cloud data of a target vehicle using multiple target grids, wherein the target vehicle point cloud includes at least one vehicle point cloud within the multiple target grids.

[0141] Optionally, the boundary determination module 1001 is further configured to: obtain the current position of the target vehicle; based on the current position, query the map information of the target vehicle from a preset map, wherein the preset map is a preset high-precision map; and extract the boundary polygon corresponding to the target vehicle from the map information, wherein the boundary polygon is determined by multiple boundary plane points.

[0142] Optionally, the region extraction module 1002 is further configured to: calculate the convex hull range corresponding to the boundary polygon according to a preset convex hull algorithm; determine a preset range of interest based on the maximum scanning range of the solid-state radar of the target vehicle; and determine the target region of interest of the target vehicle based on the intersection area of ​​the convex hull range and the preset range of interest.

[0143] Optionally, the raster extraction module 1003 is further configured to: establish a raster network based on the target region of interest of the target vehicle, wherein the raster network contains multiple candidate rasteres; calculate the region coordinate span based on the target region of interest of the target vehicle to determine the main direction corresponding to the target region of interest; and extract multiple target rasteres located within the target region of interest from multiple candidate rasteres according to the main direction and the outer contour polygon of the target region of interest.

[0144] Optionally, the grid extraction module 1003 is further configured to: initialize the bird's-eye view grid corresponding to the target vehicle based on the target interest region of the target vehicle, and establish a grid network, wherein the grid network contains multiple candidate grids as multiple candidate sector grids.

[0145] Optionally, the grid extraction module 1003 is further configured to: determine the outer contour polygon based on the outer contour point set of the target region of interest; perform scanning calculations using a preset scanning method based on the contour polygon and the main direction to obtain the calculation results; and extract multiple target sector grids located within the target region of interest from multiple candidate sector grids according to the calculation results.

[0146] Optionally, the point cloud selection module 1004 is further configured to: acquire candidate vehicle point cloud data of the target vehicle, wherein the candidate vehicle point cloud data includes multiple candidate vehicle point clouds; fill the multiple candidate vehicle point clouds into the target grid according to a preset clustering algorithm to obtain the filling result; and select the target vehicle point cloud from the multiple candidate vehicle point clouds based on the filling result.

[0147] Optionally, the point cloud selection module 1004 is further configured to: read the original data frame of the vehicle-mounted point cloud from the solid-state radar of the target vehicle; convert the point cloud coordinates in the original data frame of the vehicle-mounted point cloud to the vehicle body coordinate system of the target vehicle; and perform point cloud filtering on the original data frame of the vehicle-mounted point cloud to obtain candidate vehicle-mounted point cloud data.

[0148] According to one embodiment of the present invention, a vehicle is also provided, including an on-board memory and an on-board processor. The on-board memory stores a computer program, and the on-board processor is configured to run the computer program. When the computer program is executed by the processor to extract the region of interest on-board point cloud, it can implement the steps in any of the above-described method embodiments.

[0149] Optionally, in this embodiment, the on-board processor can be configured to perform the following steps via a computer program:

[0150] Step S1: Based on the current location of the target vehicle and the preset map, determine the boundary polygon corresponding to the target vehicle, wherein the boundary polygon is used to describe the road range where the target vehicle is located.

[0151] Step S2: Extract the target region of interest of the target vehicle based on the boundary polygon and the preset region of interest, wherein the preset region of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle.

[0152] Step S3: Extract multiple target grids located within the target region of interest from the grid network of the target vehicle;

[0153] Step S4: Using multiple target grids, select a target vehicle point cloud from the candidate vehicle point cloud data of the target vehicle, wherein the target vehicle point cloud includes at least one vehicle point cloud within the multiple target grids.

[0154] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0155] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0156] In the several embodiments provided by this invention, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be 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 system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some interfaces; the indirect coupling or communication connection of units or modules can be electrical or other forms.

[0157] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0158] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0159] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0160] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for extracting vehicle-mounted point clouds of regions of interest, characterized in that, include: Obtain the current location of the target vehicle; Based on the current location, the map information of the target vehicle is obtained from a preset map, wherein the preset map is a preset high-precision map; From the map information, the boundary polygons corresponding to the road, intersection, and tunnel entrance where the target vehicle is currently located are extracted. The boundary polygons are determined by multiple boundary plane points, which are simulated points representing each object in the preset high-precision map. The boundary polygons are used to represent the corresponding geometric figures of the road, intersection, and tunnel entrance where the target vehicle is currently located. Merge the boundary polygons corresponding to the road, intersection, and tunnel entrance where the target vehicle is currently located; Based on the merged boundary polygon and the preset range of interest, the target region of interest of the target vehicle is extracted, wherein the preset range of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle; Based on the target region of interest of the target vehicle, the bird's-eye view grid corresponding to the target vehicle is initialized to establish a grid network. The grid network includes multiple candidate grids that are multiple candidate fan-shaped grids, and the range of the grid network is the smallest circle that can cover the target region of interest. Extract multiple target sector grids located within the target region of interest from the grid network of the target vehicle; Read the original point cloud data frame from the solid-state radar of the target vehicle; The point cloud coordinates in the original vehicle point cloud data frame are transformed to the vehicle body coordinate system of the target vehicle, and then the point cloud coordinates in the original vehicle point cloud data frame are transformed from the vehicle body coordinate system to the extreme value coordinate system. The original vehicle point cloud data frame is denoised, filtered, and invalid points are removed to obtain candidate vehicle point cloud data of the target vehicle, wherein the candidate vehicle point cloud data includes multiple candidate vehicle point clouds. According to the preset clustering algorithm, the multiple candidate vehicle point clouds are filled into the target sector grid to obtain the filling result; Based on the filling result, a target vehicle point cloud is selected from the plurality of candidate vehicle point clouds, wherein the target vehicle point cloud includes at least one vehicle point cloud within the plurality of target fan-shaped grids.

2. The method according to claim 1, characterized in that, Based on the boundary polygon and the preset region of interest, the target region of interest for the target vehicle is extracted. Calculate the convex hull range corresponding to the boundary polygon according to the preset convex hull algorithm; The preset range of interest is determined based on the maximum scanning range of the solid-state radar of the target vehicle; The target region of interest for the target vehicle is determined based on the intersection of the convex hull range and the preset region of interest.

3. The method according to claim 1, characterized in that, The method further includes: Based on the target vehicle's target region of interest, the region coordinate span is calculated to determine the main direction corresponding to the target region of interest; Based on the main direction and the outer contour polygon of the target region of interest, extract the multiple target sector grids located within the target region of interest from the multiple candidate grids.

4. The method according to claim 3, characterized in that, Extracting the plurality of target sector grids located within the target interest region from the plurality of candidate grids based on the main direction and the outer contour polygon of the target interest region includes: The outer contour polygon is determined based on the set of outer contour points of the target region of interest; Based on the outline polygon and the main direction, a scanning calculation is performed using a preset scanning method to obtain the calculation result; Based on the calculation results, the target sector grids located within the target interest region are extracted from the multiple candidate sector grids.

5. A device for extracting region-of-interest (ROI) vehicle-mounted point clouds, characterized in that, include: The boundary determination module is used to calculate the current position of the target vehicle based on the global positioning information received by the global positioning system receiver and the inertial navigation data received by the inertial navigation system. Based on the current location, the map information of the target vehicle is obtained from a preset map, wherein the preset map is a preset high-precision map; from the map information, the boundary polygons corresponding to the road, intersection, and tunnel entrance where the target vehicle is currently located are extracted, wherein the boundary polygons are determined by multiple boundary plane points, and the boundary plane points are simulated points representing each object in the preset high-precision map, and the boundary polygons are used to represent the corresponding geometric figures of the road, intersection, and tunnel entrance where the target vehicle is currently located; The device is also used to merge the boundary polygons corresponding to the road, intersection and tunnel entrance where the target vehicle is currently located; The region extraction module is used to extract the target region of interest of the target vehicle based on the merged boundary polygon and the preset region of interest, wherein the preset region of interest is determined by the maximum scanning range of the solid-state radar of the target vehicle. The device is further configured to initialize the bird's-eye view grid corresponding to the target vehicle based on the target region of interest of the target vehicle, and establish a grid network, wherein the grid network includes multiple candidate grids which are multiple candidate fan-shaped grids, and the range of the grid network is the smallest circle that can cover the target region of interest. The grid extraction module extracts multiple target sector grids located within the target region of interest from the grid network of the target vehicle; A point cloud selection module is used to read the original vehicle-mounted point cloud data frame from the solid-state radar of the target vehicle; transform the point cloud coordinates in the original vehicle-mounted point cloud data frame to the vehicle body coordinate system of the target vehicle, and then transform the point cloud coordinates in the original vehicle-mounted point cloud data frame from the vehicle body coordinate system to the extreme value coordinate system; perform denoising, filtering, and invalid point removal on the original vehicle-mounted point cloud data frame to obtain candidate vehicle-mounted point cloud data of the target vehicle, wherein the candidate vehicle-mounted point cloud data includes multiple candidate vehicle-mounted point clouds; fill the multiple candidate vehicle-mounted point clouds into the target sector grid according to a preset clustering algorithm to obtain a filling result; select a target vehicle-mounted point cloud from the multiple candidate vehicle-mounted point clouds according to the filling result, wherein the target vehicle-mounted point cloud includes at least one vehicle-mounted point cloud within the multiple target sector grids.

6. A vehicle, comprising an on-board memory and an on-board processor, characterized in that, The vehicle-mounted memory stores a computer program, and the vehicle-mounted processor is configured to run the computer program to perform the method for extracting the region of interest vehicle-mounted point cloud as described in any one of claims 1 to 4.

Images