基于多线索概率分布图像融合的目标跟踪

A Novel Visual Object Tracking Algorithm Based on Multiple Clues Probability Distribution Image Fusion

为了克服多线索目标跟踪中固定线索权值的不足,提出采用颜色、运动历史、视觉显著性等多种概率分布图像来描述目标的观测信息,通过目标在前一帧的位置确定中心区域和周边区域,并计算不同概率分布图像基于这两个区域的直方图,即中心直方图和周边直方图.每个概率分布图像的置信度由其中心直方图和周边直方图的差异度来描述.根据不同线索的置信度,在线调节当前帧各种概率分布图像在目标位置判断中所占的权重,实现对目标的多线索融合跟踪.实验结果表明,比常用的固定线索权值的融合算法效果更好.

In order to overcome shortcoming of fixed clues weight,Different features are used to generate a set of likelihood maps for each pixel indicating the probability of that pixel belonging to foreground object or scene background.The confidence score of each likelihood map is computed based on distinction of histograms of likelihood values on the object versus values from the surrounding background region,measured from the likelihood map of the previous frame.The evidence combination framework dynamically updates the weights such that,in the fused likelihood map,discriminative foreground/background information is preserved while ambiguous information is suppressed.Experimental comparisons demonstrate the proposed method outperform the classical fixed clues Weight fusion technique.