A Research on stereo matching algorithm based on edge detection and gaussian disparity distribution model

In this paper, we present a novel stereo matching algorithm to create the dense disparity map for 3DTV processing line. We propose and analyze the Gaussian disparity distribution model first. Based on this model, we design the novel composite window for the window-based local optimization algorithm. An effective edge detection operator is first applied to get the object boundaries and avoid the detection of the texture edges on object surface. Then, we will apply different kinds of window on the pixels according to their locations. This scheme considers the features of the stereo images, so it reduces the mismatches and disparity fluctuations in the smooth area. Furthermore, the proposed algorithm sharps the object boundaries and improves the blurring effect caused by the common rectangular-window based algorithm. The experimental results demonstrate the effectiveness of the proposed algorithm.