Research of Vehicle Detection Algorithm base on feature points reverse tracking and optical flow clustering algorithm

摘要

First, this paper introduces the background of Rear Approaching Vehicle Detection Algorithm. Secondly, this paper summarized the algorithm. Finally, the paper presents a new algorithm and improve ideological and a brief description. The major improvements, including a new rear frame difference method and the combination of optical flow asymptotically vehicle detection algorithm and improved measures to reverse tracking and photo polymerization stream classes. Experiments show this algorithm has higher detection rate and low false alarm rate and better robustness.