CN-116051567-B - Point cloud ground segmentation method and related equipment based on harbor scene ground characteristics

CN116051567BCN 116051567 BCN116051567 BCN 116051567BCN-116051567-B

Abstract

The invention provides a point cloud ground segmentation method and related equipment based on the ground characteristics of a harbor scene. The method comprises the steps of obtaining laser point cloud data collected by a laser sensor, dividing the laser point cloud data into a plurality of grids in the horizontal plane direction, executing a ground laser point cloud data identification step on the laser point cloud data in each grid, wherein the ground laser point cloud data identification step comprises the steps of carrying out plane fitting on the laser point cloud data in the grids, responding to the laser point cloud data in the grids to be fitted into the planes and accord with the ground characteristics, determining the laser point cloud data fitted into the planes in the grids to be the ground laser point cloud data, responding to the laser point cloud data in the grids not to be fitted into the planes or accord with the ground characteristics, adjusting the size of the grids, and executing the ground laser point cloud data identification step on the laser point cloud data in the grids again. The method and the device improve the accuracy of the point cloud ground segmentation in the port scene.

Inventors

ZHANG BO
ZHANG RONG
RAO BINGBING

Assignees

上海西井信息科技有限公司

Dates

Publication Date: 20260508
Application Date: 20221116

Claims (8)

1. The point cloud ground segmentation method based on the ground characteristics of the harbor scene is characterized by comprising the following steps of: acquiring laser point cloud data acquired by a laser sensor; dividing the laser point cloud data into a plurality of grids in a horizontal plane direction; performing a ground laser point cloud data identification step on the laser point cloud data in each grid, the ground laser point cloud data identification step comprising: performing plane fitting on the laser point cloud data in the grid; Responding to the fitting of the laser point cloud data in the grids to be a plane and conforming to the ground characteristics, and determining the laser point cloud data fitted to be the plane in the grids as ground laser point cloud data; The method comprises the steps of determining the laser point cloud data in the grids, which are positioned on the same plane with the ground laser point cloud data of the adjacent grids, as quasi-ground laser point cloud data, adjusting the size of the grids, and executing the ground laser point cloud data identification step on the laser point cloud data in the grids again according to the quasi-ground laser point cloud data and the ground laser point cloud data obtained after adjustment.
2. The point cloud ground segmentation method based on the ground characteristics of the harbor scene according to claim 1, wherein the plurality of grids are divided to be the same size.
3. The point cloud ground segmentation method based on the port scene ground characteristics according to claim 2, wherein the dividing the laser point cloud data into a plurality of grids in the horizontal plane direction comprises: and determining the initial size of the grid according to the width of the vehicle body and/or the width of the road surface where the laser sensor is positioned.
4. The method for point cloud ground segmentation based on the ground characteristics of the harbor scene according to claim 2, wherein the adjusting the size of the grid comprises: the size of the grid is reduced to one quarter of the size of the original grid.
5. The point cloud ground segmentation method based on the ground characteristics of the port scene as set forth in claim 1, further comprising: acquiring ground laser point cloud data acquired by a plurality of laser sensors; and aligning the ground laser point cloud data of the plurality of laser sensors in the common view area of the plurality of laser sensors so as to perform error calibration on the laser point cloud data acquired by the plurality of laser sensors.
6. Point cloud ground segmentation device based on harbour scene ground characteristic, characterized by comprising: The laser point cloud acquisition module is used for acquiring laser point cloud data acquired by the laser sensor; The grid dividing module is used for dividing the laser point cloud data into a plurality of grids in the horizontal plane direction; A ground identification module for performing a ground laser point cloud data identification step on the laser point cloud data in each grid, the ground laser point cloud data identification step comprising: performing plane fitting on the laser point cloud data in the grid; Responding to the fitting of the laser point cloud data in the grids to be a plane and conforming to the ground characteristics, and determining the laser point cloud data fitted to be the plane in the grids as ground laser point cloud data; The method comprises the steps of determining the laser point cloud data in the grids, which are positioned on the same plane with the ground laser point cloud data of the adjacent grids, as quasi-ground laser point cloud data, adjusting the size of the grids, and executing the ground laser point cloud data identification step on the laser point cloud data in the grids again according to the quasi-ground laser point cloud data and the ground laser point cloud data obtained after adjustment.
7. An electronic device, the electronic device comprising: A processor; a storage medium having stored thereon a computer program which, when executed by the processor, performs the point cloud ground segmentation method based on the ground characteristics of a port scene as set forth in any one of claims 1 to 5.
8. A storage medium having stored thereon a computer program which, when executed by a processor, performs the point cloud ground segmentation method based on the ground characteristics of a port scene as claimed in any one of claims 1 to 5.

Description

Point cloud ground segmentation method and related equipment based on harbor scene ground characteristics Technical Field The invention relates to the field of laser radar data processing, in particular to a point cloud ground segmentation method and related equipment based on the ground characteristics of a harbor scene. Background Lidar does not distinguish between surrounding ground or other obstacles, but outputs them together as points in three-dimensional space, so that it is necessary to distinguish by an algorithm for the backend to make a correct decision. At present, the ground segmentation method comprises the following steps: First, the point cloud is encoded in combination with the laser radar transmitter scanning characteristics, and the ground is segmented by the adjacent point feature differences. For example, by traversing the points of each wire harness of the mechanical lidar one by one, when a nearby point is found to be too high in height difference, it is considered an obstacle rather than the ground. However, this method is only directed to mechanically scanned lidar and requires a radar installation angle. For solid-state lasers, or lasers require vertical scanning to be vertically installed, the recognition effect is poor. Second, radiology. The surrounding space is divided into a plurality of sector areas according to angles and radial directions, and the centers of all the sectors are the laser radars. The ground is a ray from the side view of each sector, and is shot to infinity by the laser radar. And judging whether each point and the last ground point meet gradient and height difference threshold values from the near to the far in the ray direction so as to judge whether the point is the ground. However, this method has a chain effect, if the last point is judged to be wrong, the later point will be wrong, and the steps must be performed sequentially, so that parallel calculation cannot be performed, and the time consumption is serious. Thirdly, the purpose of point cloud ground segmentation is achieved through fusion of other sensors, such as fusion with a camera, semantic segmentation of the camera and fusion with the point cloud. However, this approach relies on other sensors and requires high demands on algorithm development and system effort. In addition, the method has the problems that 1) only the characteristics of points are considered, and the whole characteristics of all points in the whole or local area are not extracted, or the extracted characteristics are single and inflexible, and 2) the laser calibration result has great influence on the ground identification effect. If the calibration error is large or the laser jolts in the movement process, the probability of segmentation failure can be greatly increased, and 3) the method aims at single laser, and in practical application, particularly in the automatic driving field, the method has become normal by using a plurality of lasers at the same time. When a plurality of lasers are spliced, different lasers can see the same ground to generate a multi-layer ground result, so that ground segmentation failure based on splicing borrowing light is caused. Therefore, how to reduce the influence of laser calibration on ground segmentation and improve the flexibility of feature extraction and the accuracy of laser point cloud ground segmentation is a technical problem to be solved by the skilled person. Disclosure of Invention In order to overcome the defects in the prior art, the invention provides a point cloud ground segmentation method, a device, electronic equipment and a storage medium based on the ground characteristics of a harbor scene, so as to reduce the influence of laser calibration on ground segmentation and improve the flexibility of feature extraction and the accuracy of laser point cloud ground segmentation. According to one aspect of the present invention, there is provided a point cloud ground segmentation method based on ground characteristics of a harbor scene, including: acquiring laser point cloud data acquired by a laser sensor; dividing the laser point cloud data into a plurality of grids in a horizontal plane direction; performing a ground laser point cloud data identification step on the laser point cloud data in each grid, the ground laser point cloud data identification step comprising: performing plane fitting on the laser point cloud data in the grid; Responding to the fitting of the laser point cloud data in the grids to be a plane and conforming to the ground characteristics, and determining the laser point cloud data fitted to be the plane in the grids as ground laser point cloud data; And in response to the laser point cloud data in the grid not fitting to a plane or the laser point cloud data in the grid not conforming to the ground characteristics, adjusting the size of the grid, and executing the ground laser point cloud data identification step on the laser point cloud data in the gr