Research And Implementation Of Binocular Parallax Ranging System

Humans get a sense of distance through the eyes, a binocular parallax ranging system get parallax through two pictures, then find the distance information. The binocular parallax ranging system which is based on the image processing is a passive ranging technology. Its advantage is that simple-requiring devices, convenient and flexible manipulation and no contact with the target. With the development of computer science and progress of optical devices, it will play a more important role in the industrial and agricultural production as well as scientific research and military fields.The key point of the system lays on three steps: feature extraction, camera calibration and image matching, which will affect the the measurement range and precision. A improved Harris corner extraction algorithm is used to achieve the sub-pixel corner extraction, including the study and adjustment of the original parameters, the improvements of efficiency and accuracy, and the sub-pixel optimization by using binomial approximation. Then the accurate camera calibration parameters is got by introducing radial and tangential distortion into the ZHANG’s calibration model. The NCC regional matching method is used to get the sub-pixel match after comparing with the SAD and SSD, it is also made to suit for ranging system.This paper studies and implements a binocular vision ranging system based on COMS cameras. Finally a binocular parallax ranging system is built, in the measurements within 200 m, its relative errors is under 4%, and the error analysis is conducted.