基于分块的自适应多特征融合目标跟踪

摘要

基于RGB直方图的目标跟踪易受背景相似颜色的影响.针对该问题,提出一种分块的自适应多特征权值融合跟踪算法.采用K-S相似性度量准则对边缘方向直方图(EOH)和RGB直方图的特征权值进行融合,实现目标跟踪.实验结果表明,与单一的RGB或EOH特征跟踪算法相比,该算法的跟踪准确率较高,具有较强的抗遮挡能力.%Tracking algorithm based on RGB histogram is sensitive to the similarity color in background,leading to failure in tracking.In view of the above problem,a tracking algorithm based on blocking with multi-feature combination is proposed.This algorithm adopts K-S similarity measure in Edge Orientation Histogram(EOH) and RGB histogram,with adaptive combination to track object.Experimental results show that it has better performance than single feature as well as robust to occlusion.