基于2DPCA与稀疏表示的目标跟踪

Target tracking based on 2DPCA and sparse representation

为了提高目标跟踪的准确性,针对目标跟踪过程中光照变化、遮挡、姿势变化等问题,提出了基于二维主成分分析(2DPCA)与稀疏表示的目标跟踪算法。在贝叶斯框架中使用了2DPCA与L2规范化呈现快速与鲁棒的目标跟踪算法。提出了新的似然函数表示方法,同时采用增量子空间学习的方法对冗余字典进行更新,有效抑制了跟踪漂移并能处理目标遮挡问题。通过对具有挑战性的跟踪视频进行定性和定量分析,实验结果证明:跟踪方法在跟踪精度上优于传统方法。

In order to improve the accuracy of target tracking, a target tracking algorithm based on two- dimensional principal component analysis（2DPCA）and sparse representation is proposed to deal with the changes of illumination, occlusion and posture during target tracking. In the Bayesian framework, 2DPCA and L2 normalization are used to present fast and robust target tracking algorithms. A new method of likelihood function representation is proposed, and the incremental subspace learning method is used to update redundancy dictionary to effectively suppress the tracking drift and to deal with the target occlusion problem. Through qualitative and quantitative analysis on challenging tracking video, experimental results show that this method is superior to the traditional tracking method in tracking precision.