摘要
Camshift算法是一种常用的跟踪算法,使用传统的Camshift算法追踪目标时,在目标与背景色调一致,或者目标与镜头相距较远的情况下可能会跟踪失败。通过分析Camshift算法的不足,从目标初始化和阈值筛选这两方面做出了改进;将原算法中的单框圈定目标改进为双框取定目标,阈值不变改进为自动调整阈值。实验结果证明,该算法具有一定的先进性,在目标较小或者目标移动迅速的情况下也能准确地进行跟踪,提高了算法对背景的适应能力以及跟踪的成功率和准确率。
Camshift （Continuously Adaptive Meanshift） algorithm is a kind of common object tracking algorithm. Under the condition of similar object color or object far away from the lens, traditional Camshift algorithm may cause the failure of tracking. Through analyzing the shortage of the Camshift algorithm, this paper made the improvement from two aspects of initialize object and selected threshold. It proposes a method of using double box to select object and adaptive screening threshold instead of the single box and the constant threshold. The experimental results indicate that the improved Camshift algorithm has certain advancement which can accurate tracking in the condition of the smaller target or move quickly. The improved algorithm enhanced the algorithm＂s ability to adapt to the background and significantly improved the rate of success and accuracy of object tracking.
出处
《科学技术与工程》
北大核心
2014年第13期231-235,共5页
Science Technology and Engineering
基金
山西省基础研究项目自然科学基金(2013011017-3)资助
