【摘要】 稀疏编码视频目标跟踪算法对目标遮挡问题有一定的适应性,但当目标受背景杂波、光照变化等干扰时,跟踪结果将会出现漂移现象.为此,提出一种基于字典学习和模板更新的视频目标跟踪算法.该算法在构造字典时加入背景模板集,利用标签一致K-SVD方法进行字典学习,同时训练出低维字典和目标背景分类器;在稀疏编码过程中,借助粒子滤波技术,采用分类器分类结果和候选目标直方图构建整体似然模型;最后通过字典学习更新字典、分类器及目标直方图.采用标准数据库中具有挑战性的视频数据进行算法测试实验,结果表明,对于存在遮挡、背景杂波、光照变化、目标旋转和尺度变化等复杂跟踪环境下的目标跟踪,文中算法都能有效地降低跟踪结果存在的漂移现象,且具有较好的稳定性.更多还原

【Abstract】 Target tracking algorithm based on sparse coding has good performance for solving the target occlusion problem. However, when the foregrounds of targets were disturbed by the background clutter and illumination, the tracking performance would be seriously decline. We therefore proposed a novel visual tracking algorithm based on dictionary learning and template updating strategy. The background template set was considered and added in the constructed dictionary, and the low dimensional dictionary and target-background classifier were trained simultaneously by using the label consistent K-SVD dictionary learning mechanism. In the sparse coding stage, the particle filter technique was employed, and the overall likelihood of each particle was calculated by using the classification results and the target candidate histogram. Finally, the dictionary, classifier and target histogram were updated by the dictionary learning method. Numerous experiments on various challenging videos show that the proposed algorithm has better tracking performance than some benchmark methods in the scenarios with the interference of occlusion, background clutter, illumination change, target rotation and scale change.更多还原