Adaptive stereo matching based on large window
Jinfeng Zhang · Jisuanji gongcheng yu sheji · 2013
Local stereo matching is difficult to obtain the accurate support area,and error matching is easy to be caused in occluded regions and textureless regions.Aiming at this problem,a new stereo matching is presented,which uses adaptive support region selection based on large window,adaptive matching cost choice and error point processing.Firstly,a larger support window is defined,in which support area of matching points is searched adaptively through the image edge information.Then,the matching cost is selected adaptively according to the number of minimum matching blocks in the support area.Finally,the error points from the left-right consistency checking of initial parallax image is classified.Experiments show that this method can achieve a good effect,as well as for occluded regions and textureless regions.