基于特征点反向跟踪和光流聚类算法的渐进汽车检测算法  

Research of Vehicle Detection Algorithm Base on Feature Points Reverse Tracking and Optical Flow Clustering Algorithm

作　　者：曹宇[1] 

机构地区：[1]工业和信息化部电子第五研究所可靠性数据中心

出　　处：《智慧工厂》2016年第8期100-103,共4页Smart Factory

摘　　要：首先介绍和总结了后方渐进车辆检测算法,其次这篇文章提出了一种更为先进高校的算法并做了明确的解释说明,主要成就包括提出了一种新的基于反向特征点跟踪和光流聚类的渐进汽车检测算法,其中反向特征点跟踪和光流聚类的思想可以去掉大部分错误跟踪,使得特征点的跟踪误差大幅度减小,试验表明此算法具有较高的检测率、较低的误检率和较高的鲁棒性。Firstly, 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,

关 键 词：车辆检测 特征点跟踪 光流聚类 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]