3D object tracking using disparity map
Wasim Akram Khan, Dibakar Raj Pant, Bhisma Adhikari, Rasana Manandhar · 2017
Object tracking is of prime essence and forms the initial step for many applications in computer vision. It is applied in gaming, computer user interaction and many more. In this article, a stereo-camera is used to detect object which makes it a low cost solution for object tracking. It solves the following problems of object tracking - knowing what an object is, which object to track and which not to, when to start tracking and when to stop. It aims at detecting objects in a video sequence and tracking them throughout the video without prior knowledge about the objects. It computes disparity map using stereo-image pair obtained from stereo-camera. Then the disparity map is subjected to depth based segmentation to detect object blobs and the corresponding region in rectified stereo-image is our object of interest.