Moving object detection and localization using stereo vision system

Abstract

The aim of this study was to design an moving object detection and localization algorithm able to detect and localize especially humans, vehicles and planes. We focused on classical methods for cameras calibration and triangulation techniques to calculate the position of the detected objects in a stereo vision rig coordinates frame. Verification of a proper operation of the proposed algorithm was made by conducting series of experiments. Our results indicates that the algorithm detects objects accurately and the troublesome un-stationary background regions can be excluded from detection using the presented localization method.