Disparity Estimation Algorithm Based on Phase and Region Segmentation

This paper proposes a stereo matching algorithm combing wavelet transform and image segmentation.Multi-resolution phase acquired from Dual-Tree Complex Wavelet Transform(DTCWT) is used to get initial disparity map.For each scale,an equation is presented to measure the matching degree.Initial disparity map is amended into a high-resolution disparity map via color segmentation and disparity plane estimation.Experimental results show that the method is simple,and can achieve high accuracy and efficiency.