Distance calculation of an object in a stereo vision system

Abstract

The ability of a machine to capture 3D information from the real world duplicating the ability of human vision is one of the great interest in science and industry. Works on stereo vision duplicating the real world environment for machines to adapt the ability of human vision by many researchers. Stereo camera based distance calculation of an object from an observer taken in an image is performed and height of the object is also calculated. Stereo parameters of both the cameras are estimated using Camera calibration. Rectification of the images from both the camera is performed so that they are in the same plane. Disparity mapping between the left and the right images is estimated. Using the disparity map and stereo parameters of both the cameras then the 3-Dimensional (3D) view of the scene is constructed. The distance of the object is calculated using Euclidean distance by determining the centroid of the bounding box and the camera and finally height of the object is calculated. Results shows the output of the object being placed at the distance between 10 to 3 m. The average error in the distance calculation is 1.74%.