Fast model based stereo matching using SOAR

Stereo correspondence is maybe the most prominent step in extraction of three dimensional structure of a scene from two or more images taken from distinct viewpoints. The correspondence problem consists of determining the locations in each image that are projections of the same physical point in space. This paper introduces a novel model based stereo matching algorithm using System of Associative Relations (SOAR) computational model. SOAR makes use of pair-wise pixel interactions to determine the underlying structure of associations within the token. The proposed stereo correspondence algorithm utilizes feature vectors (tokens) formed by direction of derivatives which constitute SOAR feature vectors. The algorithm proposed here is simple and easily realizable in hardware. Proposed algorithm is observed to yield equivalent matching performance at lower computation cost.