Collaborative model with adaptive selection scheme for visual tracking

Abstract

Visual tracking is a challenging task since it involves developing an effective appearance model to deal with numerous factors. In this paper, we propose a robust object tracking algorithm based on a collaborative model with adaptive selection scheme. Specifically, based on the discriminative features selected from the feature selection scheme, we develop a sparse discriminative model (SDM) by introducing a confidence measure strategy. In addition, we present a sparse generative model (SGM) by combining l1 regularization with PCA reconstruction. In contrast to existing hybrid generative discriminative tracking algorithms, we propose a novel adaptive selection scheme based on the Euclidean distance as the joint mechanism, which helps to construct a more reasonable likelihood function for our collaborative model. Experimental results on several challenging image sequences demonstrate that the proposed tracking algorithm leads to a more favorable performance compared with the state-of-the-art methods.