基于非负稀疏协作模型的目标跟踪算法

Object tracking via non-negative sparsity-based collaborative model

目标跟踪是计算机视觉研究领域中一个最基本的问题.为解决在复杂场景下目标跟踪效果不佳的问题,作者搭建了一个基于非负稀疏的协作模型,该模型将非负稀疏表示的产生式模型与全局模板判别式模型相结合,并提出了基于非负稀疏协作模型的目标跟踪算法.首先对每一帧图像使用粒子滤波得到若干个候选框,然后再利用非负稀疏协作模型对每一个候选跟踪框进行评分,根据得分最高判为是跟踪目标的候选框.在多个视频序列上的实验结果表明,所提出的方法可以有效地提高目标跟踪的性能.

One of the fundamental topics of computer vision is object tracking. The performance of tracking is poor in complex environment. In this paper, we established a collaborative model which combines generative model using non-negative sparse representations with discriminative classifier based on holistic templates, and proposed a tracking algorithm via non-negative sparsity-based collaborative model. In each frame, a few candidates were first obtained by using particle filter. Then a score was given to each candidate by the proposed tracking model. In the end, the candidate with the highest score was taken as the tracking result. Experiments showed that the proposed algorithm could effectively improve the performance of object tracking.