A Simple Automatic Registration Method of LiDAR Point Cloud and Optical Image

Abstract

The invention specifically discloses a simple automatic registration method of laser radar point cloud and optical image, including: preprocessing process, point cloud projection based on affine motion model, corner point feature extraction and matching, and solution through direct linear equation transformation The mapping relationship between the three-dimensional point cloud space coordinates and the optical shadow pixel coordinates is used for data fusion; the invention has the beneficial effects: 1) simple and easy to implement, and stable algorithm; 2) high degree of automation while maintaining accuracy.

Description

technical field [0001] The invention belongs to the field of laser radar point cloud and image fusion, in particular to a simple automatic registration method of laser radar point cloud and optical image. Background technique [0002] LiDAR (Light Detection and Ranging, LiDAR) is widely used in driverless cars, indoor positioning and mapping and other fields, and its development prospects are very broad. However, it is difficult to obtain target spectral information from laser point cloud data, and the color is single, which is not conducive to processing and It is understood that optical image data contains rich spectral texture and color information, which can quickly identify the attributes of ground objects and achieve better visual effects. By registering and merging 3D laser point cloud data with 2D optical image data, textured optical 3D can be obtained. The point cloud can enhance the ability to visually discriminate the attributes of the ground objects; and the regi...

AI Technical Summary

Problems solved by technology

[0002] LiDAR (Light Detection and Ranging, LiDAR) is widely used in the fields of unmanned vehicles, indoor positioning and mapping, etc., and its development prospects are very broad. However, it is difficult to obtain target spectral information from laser point cloud data, and the color is single, which is not conducive to processing and It is understood that optical image data contains rich spectral texture and color information, which can quickly identify the attributes of ground objects, and the visual effect is better. The registration and fusion of 3D laser point cloud data and 2D optical image data can obtain textured optical 3D point cloud, to strengthen the ability to visually distinguish the attributes of ground objects; while the registration of a single image and point cloud is mostly based on the method of image registration, first project the point cloud to obtain the point cloud projection image, and then use an appropriate quantification method to The projection image is processed to obtain a quantized image, and then the quantized image is registered with the optical image to obtain the index relationship between the image and the point cloud; most current automatic registration methods choose artificial calibration scenes to achieve high-precision automatic registration, which is not flexible enough.

Images