Research of Object Recognition and Tracking Based on Feature Matching

Abstract

Object recognition and tracking are very important task in several computer vision applications in our life. Most of feature matching approaches have problems which are high computational complexity and weak robustness in various environments. In this paper, we proposed a low complexity and robust object recognition and tracking using advanced feature matching for real time environment. Our algorithm recognizes object using invariant features and reduces dimension of feature descriptor to deal with the problems. Our experiments demonstrate that our work is more fast and robust than the traditional methods and can track object accurately in various environments.