Parametric contour tracking using unscented Kalman filter

This paper presents an efficient method to integrate various spatial-temporal constraints to regularize the contour tracking. The global shape of the contour is represented in a parametric form. Based on the parametric shape prior, a causal smoothness constraint can be developed. The causality nature of the constraint allows us to do efficient probabilistic contour detection using the powerful Hidden Markov Model (HMM). The contour parameters are then updated according to the detected contour points and the object dynamics by an Unscented Kalman filter (UKF). Due to the comprehensive spatial-temporal constraints, the algorithm is very robust to severe distractions. Real-time performance is also achieved. To validate the efficacy and robustness of the proposed approach, we apply this approach to track people in bad illumination and cluttered environments and promising results are reported.

• Type: Conference
• Year: 2002
• Where: ICIP
• Authors: Yunqiang Chen, Thomas S. Huang, Yong Rui
• Keywords: Contour Parameters, Contour Points, Efficient Probabilistic Contour, ICIP 2002, Image Processing