Computer vision stereo matching method

Abstract

The invention belongs to the technical field of stereo matching algorithms. The invention relates to a stereo matching method of computer vision. The method comprises the following steps: S1, giving acertain point on an image through region correlation matching, selecting a sub-window in the neighborhood of the pixel point, searching a sub-image most similar to the sub-window image in one regionof another image according to a certain similarity judgment basis, and taking the corresponding pixel point in the matched sub-image as the matching point of the pixel. According to the invention, thealgorithm provided by the invention is low in calculation complexity; and the characteristics of edge information are reserved by utilizing an infrared and gradient difference comprehensive algorithm, noise interference is reduced, the image resolution is improved, and the error parallax value of the non-stationary area is estimated by using the parallax value of the stationary area, so the mismatching rate of stereo matching can be effectively reduced, the parallax image precision is improved, and particularly, the matching performance in the area with rich edge information is better.

Description

technical field [0001] The invention relates to the technical field of stereo matching algorithms, in particular to a computer vision stereo matching method. Background technique [0002] The current level of computer stereo vision is still far from the level of human binocular vision, and its research is still a very active field. A large number of researchers in schools, companies and research institutions are still conducting research on computer stereo vision, because computer stereo vision has the following irreplaceable advantages compared with human stereo vision: 1. High precision. Although the stereoscopic perception ability of the human eye is very strong, it can easily and correctly judge the depth order of two objects, but it cannot obtain its precise distance information. However, using computer stereo vision, through accurate calibration and using appropriate calculation methods, very accurate numerical information such as depth and position can be obtained un...

AI Technical Summary

Problems solved by technology

[0004] Disadvantages of existing local matching algorithms: this method has low computational complexity and is suitable for real-time stereo matching tasks, but its disparity results are greatly affected by the size of the window and the accuracy is low

Disadvantages of existing global matching algorithms: this method has high computational complexity and is suitable for stereo matching tasks with high precision requirements. Graph cut method, belief propagation algorithm, dynamic programming, etc. are all common and efficient global stereo matching algorithms

Images