Analysis of real-time performance of stereo matching algorithm vision
Liping Li · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
Stereo Matching is one of the most important parts of achieving stereo vision system. Different three-dimensional visual systems have different requirements for matching algorithms, especially real-time stereo vision system having high performance requirements for time. Many three-dimensional matching algorithms already exist, and algorithms based on regional matching will be used to obtain a three-dimensional density disparity map. It can be applied to match the higher accuracy of the occasion, but measure of similarity needs an enormous amount of computation, which will delay the match time. If physical hardware is used to achieve matching algorithms, it can effectively improve the performance of real-time system. Regular structure algorithm, which is achieved in hardware circuits with parallel computing function, can have a higher time performance. Comparing several algorithms in real-time performance is the essential evidence to develop real-time visual systems.