A New Fast Matching Approach of Large Disparity Stereopair

A new fast area-based stereo algorithm suitable to large disparity stereo pair matching and real time applications is proposed in this paper. Firstly, the proposed approach estimates the global disparity of stereo pairs and gets two maximum correlation areas of left and right images in estimation disparity level. Then, matching algorithm based on area correlation is used to match in the two maximum correlation areas based on the global estimation disparity, the computation optimization strategy is applied to speed the algorithm, the sum of disparity computed in maximum correlation areas and the global estimation disparity is the result disparity. Finally, this paper provides the compared experiments. The results indicate that combining global estimation disparity strategy with fast calculation scheme is valid and feasible.