Adaptive Appearance Tracking Model Using Subspace Learning Method

Visual tracking is still a challenging subject due to the targeted object’s change in direction and size, stochastic disturbance under complicated scene. In the work, we proposed a visual tracking framework based on the subspace’ updating and learning. We introduced the Hall’s subspace updating algorithm and the new measurement on subspace’s similarity in computing particles’ weights under Condensation algorithm in our tracking processes. Differed from conventional PCA method, our method adaptively updated the subspace which can reflect appearance variation of the moving target over long period of time. Compared with Condensation algorithm using color histogram, the tracker we proposed can effectively track the target under complicated surrounding.