A phase matching and region segmentation stereo depth mixed extraction algorithm

This paper presents a novel dense disparity estimation method using phase matching and color segmentation. In the proposed method, the initial disparity map is firstly obtained by an improved phase-based stereo matching solution. The solution effectively restrains two tough problems of phase singularities and phase wrap, in a coarse-to-fine approach, by using Dual-Tree Complex Wavelet Transform (DT CWT) to extract the phase of the stereo images. The initial disparity is then refined by color segmentation and disparity plane fitting. Segmentation-based methods have an advantage of well representation of disparities on discontinuous region. Experimental results indicate that the proposed method is superior in generating disparity maps with high accuracy and clean edges.