Window-based Stereo Matching Algorithm Using a Weighted Average of Costs Aggregated with Window Size Reduction

In the intensity-based stereo matching algorithm, window-based and coarse-to-fine algorithms are known as typical approaches. The window-based algorithm matches intensity values within windows between two stereo images. The conventional window-based algorithms, however, face a trade-off between error rates of the disparity map in disparity continuity and discontinuity regions due to the window size dependence. To solve the problem, the coarse-to-fine algorithm has been proposed [2-3]. The algorithm starts the matching process by the largest window size and gradually decreases the window size with narrowing a rage of candidates. The algorithm, however, often cannot find true disparities due to a limitation of candidates of disparity. To solve these issues, we proposed a new algorithm of the window-based stereo matching.