Visual tracking based on the color attention preserved sparse generative object model

Abstract

A new color attention preserved sparse generative object model is proposed to handle occlusion and illumination variations in the visual tracking task. The color attention is represented by the fast calculated color descriptor on color names, which is used to weight the similarity measurement of the sparse generative model. In the sparse generative model, the image region of the object is divided into patches. And we take into consideration of the spatial information of each patch to handle the occlusion. The color attention assists the sparse generative model to match the spatial structure of the object and handle illumination variation and heavy occlusion on the object. The update scheme considers both the original template and the latest observations, which enables the tracker to deal with appearance change effectively and alleviate the drifting problem. Experiments on the several video databases under the influence of various factors demonstrate the efficiency of the proposed method.