Monocular vision-based dense point cloud reconstruction method and system for triangulation measurement depth

Abstract

The invention provides a monocular vision-based dense point cloud reconstruction method and system for triangulation depth measurement. The monocular vision-based dense point cloud reconstruction method comprises the steps of reading an image in a data set; matching the pixel points in the read front and back frame images by adopting an epipolar search and block matching algorithm; calculating thedepth value of the matched pixel point through a triangulation principle; performing filtering processing on the depth values of the pixel points by adopting a depth filter in Gaussian distribution,and keeping the depth value of the pixel point with the minimum uncertainty; segmenting the image into 4 * 4 pixel blocks; if the depth value difference of each pixel point in each pixel block does not exceed a set threshold value, regarding the whole pixel block as a point cloud point; and if the depth value difference of each pixel point in each pixel block exceeds the set threshold value, regarding the pixel points with the maximum depth value and the minimum depth value in the block as two point cloud points; converting the pixel coordinates into world coordinates by using a camera model,and generating point cloud data in combination with the depth information. The problem that conventional point cloud map creation needs multiple sensors or expensive sensors is solved.

Description

technical field [0001] The present invention relates to the technical field of unmanned driving environment perception, in particular, to a dense point cloud reconstruction method and system based on monocular vision triangulation measurement depth. Background technique [0002] At present, the creation of point cloud maps is basically based on the combination of laser sensors and monocular cameras to directly measure the depth map and color map, and then use the camera model to obtain point cloud maps according to the sensor pose, or use methods such as multi-sensor fusion. , the algorithm architectures adopted by these methods are relatively simple and direct, which will lead to insufficient image processing, insufficient processing of edge information or image areas with large differences in depth values, and the cost of multi-sensor fusion is relatively expensive. The dense point cloud reconstruction method based on monocular vision triangulation measurement depth can so...

AI Technical Summary

Problems solved by technology

[0002] At present, the creation of point cloud maps is basically based on the combination of laser sensors and monocular cameras to directly measure the depth map and color map, and then use the camera model to obtain point cloud maps according to the sensor pose, or use methods such as multi-sensor fusion. , the algorithm architecture adopted by these methods is relatively simple and direct, so it will lead to insufficient image processing, insufficient processing of edge information or image areas with large differences in depth values, and the cost of multi-sensor fusion is relatively expensive

Images