计算机应用与软件2011，Vol.28Issue(10)：72-76,5.
面向交通流检测的Mean Shift多目标自适应跟踪算法
TRAFFIC FLOW DETECTION-ORIENTED MEAN SHIFT MULTI-TARGET ADAPTIVE TRACKING ALGORITHM
摘要
Abstract
With the development of the society, traffic problems are increasingly serious. Traffic flow detection technology is an important means to solve this problem, and target tracking is an essential part in detection technology. Mean Shift algorithm is a widely used technology now in the field of target tracking, but the computation of traditional Mean Shift tracking algorithm is so big, and is very difficult to achieve the real-time multi-target tracking in traffic flow detection. In view of this, the paper proposes a new Mean Shift tracking algorithm based on linear prediction. The algorithm introduces a prediction vector for predicting the possible location the target might occur in the next frame. When tracking the target, the algorithm starts the iteration from predictive position until converging at the true location of the target. Experimental results show that the algorithm improves the efficiency of original algorithm to a great extent, and benefits the real-time tracking. Besides, to solve the adaptive bandwidth issue of kernel function in Mean Shift tracking algorithm, the article proposes a new method to realise adaptive bandwidth update based on comparing the proportion variation in target centre grey degree.关键词
交通流/Mean Shift/直方图/预测矢量/自适应Key words
Traffic flow/Mean Shift/Histogram/Predictive vector/Adaptive分类
信息技术与安全科学引用本文复制引用
闫德莹,刘贵全,缪泓..面向交通流检测的Mean Shift多目标自适应跟踪算法[J].计算机应用与软件,2011,28(10):72-76,5.基金项目
国家高技术研究发展计划(2009AA01Z132). （2009AA01Z132）
