Stereo images for depth measurement

Abstract

Obtaining 3D information from 2D image/video or a pair of 2D images is a way more economical and handy than that of using 3D cameras. The depth by disparity approach to obtain 3D information from a pair of 2D images called as stereo images is used in said system. Though the concept of depth by disparity is not new, the significant progress and useful applications has not been yet seen in the real world. The system proposed presents a method/algorithm to extract the disparity map (Depth information) within milliseconds by using a pair of stereo images of that particular scene. The time taken to calculate disparity map is 25 and 30 seconds using combined SAD and GRAD algorithm. To implement system in real time, the execution time needs to be very small. Thus a modified algorithm is presented in this paper to minimize the time taken to compute the disparity map by minimizing the calculations. The time saving is achieved by eliminating the unwanted search regions (which minimizes active search region) for disparity computation.