A system on chip based stereo vision approach for disparity measurement

摘要

Stereo vision systems are very inevitable part for depth measurement in most autonomous systems. This paper presents an approach to stereo vision applicable for finding disparity and depth information which is very critical for autonomous robot navigation[1] and path following applications. Using the depth information[2] it is possible to control drones and similar autonomous systems. A pair of USB webcams was used and were interfaced with Raspberry Pi, which is a Linux based low powered system on chip which makes the system very compact. The whole disparity measurement system is capable of calibrating the stereo camera rig using a light weight cross platform and powerful tool called Qt creator. The Qt creator API is linked with the dedicated real time image processing library OpenCV and for this project OpenCV[3] functions are used exclusively for calibration and stereo vision disparity measurement.