CN-116953727-B - Object detection method, device, electronic equipment and storage medium

CN116953727BCN 116953727 BCN116953727 BCN 116953727BCN-116953727-B

Abstract

The application relates to an object detection method, an object detection device, electronic equipment and a storage medium. The object detection method comprises the steps of obtaining first point cloud data obtained by scanning a lane area corresponding to a vehicle, obtaining first path edge data obtained by performing path edge detection on the first point cloud data, performing amplification processing on the first path edge data to obtain second path edge data, determining second point cloud data located in the lane area in the first point cloud data based on the second path edge data, clustering based on the second point cloud data to obtain candidate objects, determining a target area in the lane area based on the first path edge data, and determining the candidate objects in the target area as target objects to be detected. The method and the device can filter out the interference data of the area outside the lane, realize the filtering of the virtual target of the area inside the lane, filter out the candidate object outside the target area and improve the accuracy of the detection of the target object.

Inventors

PENG HUILING
LU YIXIN
Chen Wangshuangyi
DENG HAOYUN
QIAN SHAOHUA

Assignees

重庆长安汽车股份有限公司

Dates

Publication Date: 20260508
Application Date: 20230727

Claims (11)

1. An object detection method, comprising: acquiring first point cloud data obtained by scanning a lane area corresponding to a vehicle; Acquiring first path edge data obtained by performing path edge detection on the first point cloud data, and performing amplification processing on the first path edge data to obtain second path edge data; the first path edge data comprises a plurality of path edge points, and the first path edge data is amplified to obtain second path edge data, which comprises the following steps: Inserting a road edge point for filling and complementing into the first road edge data to obtain complementing road edge data, so that each road edge point in the complementing road edge data corresponds to a corresponding resolution angle in a scanning visual field range; Inserting invalid road edge points for occupying positions into the full road edge data to obtain second road edge data, so that each resolution angle in the scanning view field range has corresponding road edge points in the second road edge data; Determining second point cloud data of an area in a lane from the first point cloud data based on the second road edge data, and clustering to obtain candidate objects based on the second point cloud data; And determining a target area in the lane area based on the first road edge data, and determining a candidate object in the target area as a target object to be detected.
2. The method according to claim 1, wherein inserting the first edge data into the edge points for filling and complementing to obtain the complemented edge data, comprises: interpolation processing is carried out between the road edge points with the intervals smaller than or equal to a first threshold value in the first road edge data, and interpolation road edge data are obtained; And performing complement processing on the interpolation road edge data between the road edge points with intervals larger than the first threshold and smaller than the second threshold to obtain complement road edge data.
3. The method of claim 1, wherein inserting invalid edge points for occupying positions into the full edge data to obtain second edge data comprises: And inserting invalid road edge points for occupying positions between the road edge points with the interval larger than or equal to a second threshold value in the complement road edge data to obtain second road edge data.
4. The object detection method according to claim 1, wherein determining second point cloud data located in an in-lane region among the first point cloud data based on the second road edge data includes: acquiring coordinates of each point in the first point cloud data, and determining a resolution angle corresponding to each point; calculating a first polar distance of coordinates of a point corresponding to each resolution angle by taking the position of the vehicle as an origin; determining a first road edge distance threshold corresponding to each resolution angle based on the second road edge data; if any one of the first polar distance is smaller than or equal to the corresponding first road edge distance threshold value, determining that the point is positioned in the lane line; And determining the coordinates of all the points positioned in the lane line as second point cloud data of the area positioned in the lane.
5. The object detection method of claim 4, wherein determining a first edge distance threshold for each resolution angle based on the second edge data comprises: Determining effective road edge points except for the ineffective road edge points in the second road edge data, wherein the resolution angle corresponds to each effective road edge point; calculating a first distance between the position of the vehicle and each effective road edge point to serve as a first distance threshold; for invalid road edge points in the second road edge data, determining a resolution angle corresponding to each invalid road edge point; determining corresponding default route edge points in default route edge boundaries based on the resolution angle; calculating a second distance between the position of the vehicle and each default route edge point to serve as a second distance threshold; And determining a first road edge distance threshold corresponding to each resolution angle by the first distance threshold corresponding to each resolution angle and the second distance threshold.
6. The object detection method according to claim 1, wherein determining a target area in the lane area based on the first road edge data comprises: Acquiring coordinates of a first road edge point which is positioned on the left side of the position of the vehicle and is farthest from the position of the vehicle, and coordinates of a second road edge point which is positioned on the right side of the position of the vehicle and is farthest from the position of the vehicle from the first road edge data; And determining the target area based on the coordinates of the first road edge point and the coordinates of the second road edge point.
7. The object detection method according to claim 1, wherein determining the candidate object within the target area as the target object to be detected includes: Acquiring coordinates of the candidate objects, and determining a resolution angle corresponding to each candidate object; Calculating a second path distance of the candidate object corresponding to each resolution angle by taking the position of the vehicle as an origin; determining a second road edge distance threshold corresponding to each resolution angle based on the target area; If any one of the second path distances is smaller than or equal to a corresponding second path distance threshold, determining that the candidate object is located in the target area; the target object is determined based on the candidate objects located within the target area.
8. The object detection method according to claim 7, wherein determining as the target object based on the candidate objects located within the target area includes: Acquiring the size characteristics of each candidate object; and determining the candidate object with the size characteristic being greater than or equal to a preset size threshold as the target object.
9. An object detection apparatus, comprising: The first acquisition module is used for acquiring first point cloud data obtained by scanning a lane area corresponding to a vehicle; The second acquisition module is used for acquiring first path edge data obtained by performing path edge detection on the first point cloud data, and performing amplification processing on the first path edge data to obtain second path edge data; the first path edge data comprises a plurality of path edge points, and the first path edge data is subjected to amplification processing to obtain second path edge data, wherein the first path edge data is inserted with the path edge points used for filling and complementing to obtain complementing path edge data, so that each path edge point in the complementing path edge data corresponds to a corresponding resolution angle in a scanning view field range; The first determining module is used for determining second point cloud data of an area in a lane in the first point cloud data based on the second path edge data and clustering to obtain candidate objects based on the second point cloud data; And the second determining module is used for determining a target area in the lane area based on the first path edge data and determining a candidate object in the target area as a target object to be detected.
10. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus; A memory for storing a computer program; A processor, configured to implement the object detection method according to any one of claims 1 to 8 when executing a program stored in a memory.
11. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a program of an object detection method, which when executed by a processor, implements the steps of the object detection method according to any of claims 1-8.

Description

Object detection method, device, electronic equipment and storage medium Technical Field The present application relates to the field of autopilot, and in particular, to an object detection method, apparatus, electronic device, and storage medium. Background In the development of an automatic driving system, object detection on an obstacle is always the most basic module in a perception stage, wherein point cloud information received by a laser radar sensor is important basic information for object detection. The laser radar has the point cloud information of hundreds of thousands of quantity per frame, not only comprises the information of target vehicles in the road, but also can describe the whole environment of the road in detail, including pedestrians, and other obstacles. In the target detection of the laser radar point cloud with rich information, the division of the road area is always a hot spot problem, and the number of non-main targets output to the downstream module is reduced mainly for dividing the main targets and the sub-targets according to the inside and the outside of the lane, so that the complexity of the downstream module on target processing is reduced, and unnecessary expense of target processing is reduced. The most common road zone divisions are generally based on default zone ranges or forward vision lane lines. The default area range is not a dynamic updated region of interest (Region of Interest, ROI), can not adapt to various road conditions, is used for target filtering, and always has the problems of under-filtering and over-filtering targets, and the forward vision visual lane line uses a three-lane model to distinguish the target detection range from the inside of the road and the outside of the road, so as to filter the targets. However, when the visual three-lane model is applied to target detection and filtration of laser radar point clouds, although a plurality of secondary targets outside the road can be filtered, the laser radar point clouds can lose much environmental information, such as intersection scenes, forward-looking lane lines cannot be effectively identified, the lane line distance is too short, and a large number of targets in the lane are filtered, so that detection omission is caused to a certain extent. Disclosure of Invention In order to solve the technical problems described above or at least partially solve the technical problems described above, the present application provides an object detection method, an object detection device, an electronic device, and a storage medium. In a first aspect, the present application provides an object detection method, including: acquiring first point cloud data obtained by scanning a lane area corresponding to a vehicle; Acquiring first path edge data obtained by performing path edge detection on the first point cloud data, and performing amplification processing on the first path edge data to obtain second path edge data; Determining second point cloud data of an area in a lane from the first point cloud data based on the second road edge data, and clustering to obtain candidate objects based on the second point cloud data; And determining a target area in the lane area based on the first road edge data, and determining a candidate object in the target area as a target object to be detected. Optionally, the first path edge data includes a plurality of path edge points, and the amplifying process is performed on the first path edge data to obtain second path edge data, including: Inserting a road edge point for filling and complementing into the first road edge data to obtain complementing road edge data, so that each road edge point in the complementing road edge data corresponds to a corresponding resolution angle in a scanning visual field range; And inserting invalid road edge points for occupying positions into the full road edge data to obtain second road edge data, so that each resolution angle in the scanning view field range has corresponding road edge points in the second road edge data. Optionally, inserting the first edge data into the edge points for filling and complementing to obtain complementing edge data, including: interpolation processing is carried out between the road edge points with the intervals smaller than or equal to a first threshold value in the first road edge data, and interpolation road edge data are obtained; And performing complement processing on the interpolation road edge data between the road edge points with intervals larger than the first threshold and smaller than the second threshold to obtain complement road edge data. Optionally, inserting an invalid edge point for occupying a space into the complement edge data to obtain second edge data, including: And inserting invalid road edge points for occupying positions between the road edge points with the interval larger than or equal to the second threshold value in the complement road edge data to obtain second road edge