Rectangular subregioning and 3-D maximum-surface techniques for fast stereo matching

This paper presents a fast and reliable stereo matching algorithm which produces a dense disparity map by using fast cross correlation, rectangular subregioning and 3D maximum-surface techniques in a coarse-to-fine scheme. Fast correlation is achieved by using the box filtering technique whose speed is invariant to the size of correlation window and by segmenting the stereo images at different levels of the pyramid into rectangular subimages. The disparity for the whole image is found in the 3D correlation coefficient volume by obtaining the maximum-surface using our novel two-stage dynamic programming technique. There are two original contributions in this paper: (1) development of a rectangular subregioning (RSR) technique for fast similarity measure; and (2) development of a novel two-stage dynamic programming (STDP) technique for obtaining 3D maximum surface in a 3D volume efficiently. Typical running time of our algorithm implemented in C language on a 512×512 image is in the order of a few seconds. A variety of synthetic and real images have been tested, and good results have been obtained