A 3D Scenes Reconstruction Method Based on Less Featured Block Matching with a Binocular Vision System

Abstract

This paper presents a 3D scenes reconstruction method based on less featured block matching with a binocular vision system. The basic idea of this paper is to utilize the imaging parallax of two cameras caused by different setup locations to immediately and precisely obtain the 3D information of the objects within the field of view of the cameras. To accomplish this, we first adjust the setup locations of the cameras and then conduct the pre-processing operations on both left and right images, such as image rectification, grey-level transformation, and histogram equalization. After that, we divide the right image into blocks and adopt the Pearson product moment correlation coefficient to measure the similarity of the corresponding blocks in the left image. If the image variance of the current block is too large, we further partition this block into some sub -blocks to recursively perform the matching. We also propose an efficient searching algorithm to reduce the execution time of block matching. Once we fi nd the disparity of the corresponding blocks, we reconstruct the real 3D location by use of a stereo imaging model. The comparison of the 3D scenes reconstruction results with paralleled and non-paralleled cameras approaches is made through doing many experiments. We conclude that the paralleled cameras approach attains higher reconstruction accuracy than the non-paralleled cameras approach does although the former is a special case of the latter.