Structural similarity based disparity estimation for stereoscopic images

Depth estimation of a scene from stereo images is an active area of research in computer vision applications. In this paper, a simpler and faster technique for depth estimation, based on structural similarity index (SSIM) method, is proposed. The technique provides a simple block based similarity measure that considers the structural similarity between the corresponding blocks of the left and right views. It avoids explicit calculations in depth or disparity estimation. The method shows that the estimation performance is acceptable and it can be used in autonomous navigation and 3D media applications.