基于局部稀疏表示的多示例学习跟踪方法研究

Multiple Instance Learning Tracking Method with Local Sparse Representation

针对目标纹理变化、光照和位置变化较大时,跟踪不稳定、易丢失目标的问题,提出通过多示例学习的训练数据生成局部稀疏编码,建立对象的外观模型。首先,目标对象的局部图像块由过完备字典结合稀疏编码表示;其次,分类器学习稀疏编码进而识别背景中的目标;最后,将训练分类器得到的结果输入粒子滤波框架,进而预测目标状态随时间的变化。此外,为了减少字典更新和分类器累积误差形成的视觉漂移,采用弱分类器结合强分类器进行目标跟踪。

When objects undergo large pose change, illumination variation or partial occlusion, most existed visual tracking algo-rithms tend to drift away from targets and even fail in tracking them. To address this issue, in this study, the authors propose an on-line algorithm by combining multiple instance learning(MIL) and local sparse representation for tracking an object. The key idea inour method is to model the appearance of an object by local sparse codes that can be formed as training data for the MIL frame-work. First, local image patches of a target object are represented as sparse codes with an over complete dictionary. Then MILlearns the sparse codes by a classifier to discriminate the target from the background. Finally, results from the trained classifier areinput into a particle filter framework to sequentially estimate the target state over time in visual tracking. In addition, to decreasethe visual drift because of the accumulative errors when updating the dictionary and classifier, a two-step object tracking methodcombining a weak classifier with a strong classifier is proposed.