Stereo Image Reconstruction Using Adaptive Window Approach

Abstract

Image reconstruction is the art of obtaining a three dimensional scene from 2D images.. Approaches for image reconstruction are, Shape from shading, Shape from texture, Shape from motion, and Shape from stereo. In Shape from stereo or the Stereo Image Reconstruction only two images are used to reconstruct the scene. The most difficult problem in Stereo image reconstruction is the correspondence problem. Matching each pixel is a complex process results in many mismatches and inaccurate three dimensional scene. We have done matching in two levels, the first level is Feature based matching, the second level is Area-based matching. We have joined the two levels to obtain a more accurate pixel by pixel correspondence. We also obtained an adaptive search range and window size for each pixel to reduce the mismatches. Our procedure introduced high accuracy results and more dense 3D scene.