Stereo matching method for stereo binocular vision measurement

Abstract

The invention discloses a stereo matching method for stereo binocular vision measurement. The method comprises the following steps of: after projecting a Gray code coding picture with the least period T =16, acquiring the Gray code coding picture by using a left camera and a right camera; respectively computing the Gray code values of all pixels in a left picture and a right picture; generating a sine projection grating picture based on optimized sine grating parameters; acquiring the grating picture by using the left camera and the light camera after projecting; computing the phase values of all pixels of the left picture and the right picture; correcting the period misplacement between the Gray code values and the phase values by adopting a maximum probability correction method so that the pixel points on the edge of a Gray code period form a correct and complete phase period; and outputting all matching point data after searching corresponding matching points in the left picture and the right picture by adopting a two-step matching method. By the method, the corresponding matching points in the left picture and the right picture can be searched rapidly and accurately, and the precision for searching the corresponding points is improved.

Description

technical field [0001] The invention belongs to the technical field of reverse engineering and relates to a stereo matching method for binocular stereo vision measurement. Background technique [0002] Stereo matching in binocular stereo vision refers to finding the corresponding pixel position of the same visible point on the surface of the object in the sampling pictures of the left and right cameras. This is the most complex and important link in binocular stereo vision. In an environment where camera calibration is basically mature, the accuracy of stereo matching has become the most critical and direct factor affecting the accuracy of 3D measurement. Therefore, stereo matching has become a hot spot, difficulty, and focus for stereo vision researchers at home and abroad. At present, in the stereo matching of binocular stereo vision, technologies such as Gray code grating encoding and decoding, sinusoidal grating phase resolution, and stereo correction have been relative...

AI Technical Summary

Problems solved by technology

However, in the actual measurement process, due to the inconsistent sampling of the left and right cameras and the influence of random noise, the number of matching points found by direct matching is small, and only sparse and discontinuous disparity maps can be produced.

[0005] To sum up, there are problems such as low matching efficiency and a small number of matching points in the traditional binocular stereo vision measurement method for matching and searching the corresponding points of the left and right images.

Images