3D Reconstruction Based On Block Matching and Image Pyramiding
K. Rasool Reddy, Phani Raja, M.S. Rao, Sir C. R. Reddy · 2014
Nowadays, there is an increase in the usage of geometric 3D models in movie industry, games and virtual environments.There are several existing solutions like user-driven models, 3D scanners but they are not fully satisfying in either in terms of cost and or in efficiency. To overcome these problems, we have introduced a cost-effective solution based on analyzing image sequences. The main focus of this paper is on stereo vision, it is the process of recovering depth from camera images by comparing two or more views of the same scene.This paper mainly focuses on how to compute the depth map between two rectified stereo images.In this paper we use block matching, which is the standard algorithm for high-speed stereo vision in hardware systems. We first explore basic block matching, and then apply dynamic programming to improve accuracy, and image pyramiding to improve speed. Keywords: Images, 3D reconstruction, stereo vision, block matching, sub-pixel estimation, dynamic programming, image