CN-122017795-A - 2D laser radar point cloud data processing method

CN122017795ACN 122017795 ACN122017795 ACN 122017795ACN-122017795-A

Abstract

The invention relates to the technical field of laser radars, and particularly provides a 2D laser radar point cloud data processing method. The method comprises setting a distance threshold for data processing And echo intensity threshold Acquiring a circle of point cloud data to obtain the actual measurement distance of the radar And echo intensity Distance to reserve Less than a preset distance Will be greater than Setting the data points of the window as invalid values to obtain a new data set, judging the boundary of the target object according to the invalid values to obtain the index value of the boundary, setting a sliding window and the sliding times according to the echo intensity characteristics, comparing the average value of the echo intensities in the adjacent windows, and if the difference value is larger than the set threshold value Or the set sliding times are reached, the point cloud is reserved, the data processing is stopped, and otherwise, the edge widening points are deleted. The method effectively eliminates the problem of widening at the edge of the target object, and restores the real outline of the scanned target object.

Inventors

SHAO GUANGCUN
ZHU CHENGXIANG
LU NING
FANG XINXIN
PAN YANGYANG
RUAN QIANG
Zheng Zhaozhou
ZHANG SHUAI
LI FENG

Assignees

济宁科力光电产业有限责任公司
山东科力光电技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (9)

1. A 2D laser radar point cloud data processing method, the method comprising the steps of: S1, presetting a distance threshold D and an echo intensity threshold I for 2D laser radar point cloud data processing; S2, acquiring complete point cloud data of single-circle scanning of the 2D laser radar, and extracting an actual measurement distance D and echo intensity i corresponding to each scanning point; S3, comparing the actual measured distance D of each scanning point with a distance threshold D, reserving point cloud original data with D < D, setting the scanning points with D more than or equal to D as invalid values, and generating a single-circle filtered point cloud data set; s4, traversing the single-circle filtered point cloud data set, completing target object segmentation through invalid values, and recording boundary index values of each segmented target object; S5, based on the characteristic that the echo intensity in the object is larger than that of the edge, setting a sliding window X and a maximum sliding frequency Y by taking a boundary index value as a starting point, sliding and comparing the average value difference value of the echo intensities of the adjacent windows inwards from the boundaries at two sides of the target object, if the difference value is larger than an echo intensity threshold I or the maximum sliding frequency Y is reached, reserving residual point cloud data and terminating the processing, otherwise, deleting the corresponding edge widening point.
2. The 2D lidar point cloud data processing method according to claim 1, wherein in step S1, the distance threshold D is calibrated by taking the reflective patch test piece as a reference, the radar position is fixed, and then the reflective patch test piece is moved to the position with the minimum widening point, and the distance corresponding to the position is taken as the distance threshold D; the echo intensity threshold I is calibrated by the maximum fluctuation value of the echo intensity in the moving process of the reflective patch test piece.
3. The 2D lidar point cloud data processing method according to claim 1, wherein in step S2, the acquired single-turn point cloud data includes a scan angle corresponding to each scan point, an actual measurement distance D, and an echo intensity i, and the total number of single-turn scans is determined by a radar scan angle range and an angle resolution.
4. The 2D lidar point cloud data processing method according to claim 1, wherein in step S3, each scanning point of the single-circle point cloud is compared point by point according to a scanning angle sequence, a sequence number n of the scanning point satisfies 1 n and is less than or equal to a total number of single-circle scanning points, and filtering and invalid value assignment of the full-quantity point cloud are completed.
5. The 2D lidar point cloud data processing method of claim 1, wherein step S4 further comprises: marking the left boundary and the right boundary of the target object according to the recorded boundary index value, and simultaneously calculating the number N of effective scanning point clouds corresponding to the target object based on the boundary index value, wherein N is more than or equal to 2.
6. The 2D lidar point cloud data processing method according to claim 1, wherein in step S5, the sliding window X is the number of continuous point clouds processed once, and X is equal to or greater than 2; the maximum sliding times Y are the maximum sliding times of the window to the inside of the object, and Y is more than or equal to 2.
7. The 2D lidar point cloud data processing method according to claim 1, wherein in step S5, an average value of echo intensities of all points in a single sliding window is calculated, then an intensity average value difference Δi of two adjacent sliding windows is calculated, and if Δi is less than an echo intensity threshold I, all points in the current window are determined to be edge widening points and deleted.
8. The 2D lidar point cloud data processing method of claim 1, wherein in step S5, the sliding process logic of the left boundary and the right boundary of the target object is identical, and the sliding process is synchronized from the left boundary to the right and vice versa.
9. The 2D lidar point cloud data processing method according to claim 1, wherein in step S5, the sliding window X and the maximum sliding number Y are adaptively adjusted according to the features of near dense and far sparse of the 2D lidar point cloud.

Description

2D laser radar point cloud data processing method Technical Field The invention belongs to the technical field of laser radars, and particularly relates to a 2D laser radar point cloud data processing method. Background The 2D laser radar has high ranging precision, high scanning speed and strong environment anti-interference performance, and the contour restoration precision of the point cloud data directly determines the scene sensing and measuring accuracy. In an ideal state, the laser radar emission beam is an infinite thin straight line with a cross section which can be regarded as a geometric point, but in practical application, the laser radar emission beam is influenced by factors such as inherent divergence angle of a laser, installation deviation, error of a collimating lens surface, manufacturing tolerance of a structural member and the like, and the emergent beam can be diffused to form a light spot with a certain area. When the laser light spot irradiates the edge of the target object, weak echoes generated by part of the light spot can still be captured by the receiver, the weak echoes are identified as effective object point clouds, edge widening points are formed, and the measured object size is larger than the actual physical size. In the prior art, the conventional point cloud filtering method can only filter discrete outlier noise points and remote background points, cannot accurately remove edge widening points continuous with an object body, is easy to have the problems of false deletion of effective points and incomplete removal of widening points, and cannot restore the real contour of a target object. Therefore, a 2D laser radar point cloud data processing method capable of efficiently eliminating edge stretching is needed. Disclosure of Invention In view of the above, the present invention provides a 2D lidar point cloud data processing method for eliminating the widening points and restoring the real contour of the scanned object. The method comprises the following steps: S1, presetting a distance threshold D and an echo intensity threshold I for 2D laser radar point cloud data processing; S2, acquiring complete point cloud data of single-circle scanning of the 2D laser radar, and extracting an actual measurement distance D and echo intensity i corresponding to each scanning point; S3, comparing the actual measured distance D of each scanning point with a distance threshold D, reserving point cloud original data with D < D, setting the scanning points with D more than or equal to D as invalid values, and generating a single-circle filtered point cloud data set; s4, traversing the single-circle filtered point cloud data set, completing target object segmentation through invalid values, and recording boundary index values of each segmented target object; S5, based on the characteristic that the echo intensity in the object is larger than that of the edge, setting a sliding window X and a maximum sliding frequency Y by taking a boundary index value as a starting point, sliding and comparing the average value difference value of the echo intensities of the adjacent windows inwards from the boundaries at two sides of the target object, if the difference value is larger than an echo intensity threshold I or the maximum sliding frequency Y is reached, reserving residual point cloud data and terminating the processing, otherwise, deleting the corresponding edge widening point. Optionally, in step S1, the distance threshold D is calibrated by taking the reflective patch test piece as a reference, the radar position is fixed, and then the reflective patch test piece is moved to the position with the minimum widening point, and the distance corresponding to the position is taken as the distance threshold D; the echo intensity threshold I is calibrated by the maximum fluctuation value of the echo intensity in the moving process of the reflective patch test piece. Optionally, in step S2, the acquired single-circle point cloud data includes a scan angle corresponding to each scan point, an actual measurement distance d, and an echo intensity i, and the total number of single-circle points is determined by the radar scan angle range and the angle resolution. Optionally, in step S3, each scanning point of the single-circle point cloud is compared point by point according to the scanning angle sequence, the sequence number n of the scanning point satisfies that n is greater than or equal to 1 and less than or equal to the total number of single-circle scanning points, and the filtering and invalid value assignment of the full-quantity point cloud are completed. Optionally, step S4 further includes: marking the left boundary and the right boundary of the target object according to the recorded boundary index value, and simultaneously calculating the number N of effective scanning point clouds corresponding to the target object based on the boundary index value, wherein N is more than or equal to 2.