A Long Time Tracker Based on Self-Correction and Random Selection Strategy

Visual tracking is one of the most active research areas in computer vision, with numerous application including augmented reality, surveillance, weapons-guiding technologies, and object identification. The key question for robust visual tracking is to extract the appearance model features of target appropriately and locate the object exactly. This paper presents a novel nonrigid target tracking algorithm which can track a nonrigid target(such as hand) stably and is real time, and we review on the latter target tracking research direction, which is a very challenging problem in the field of visual tracking. Our algorithm is based on a SCTS (self-correction tracking strategy), then we use a RSHD (random selection HAAR detector) to detect the real target, the final target location is the mixed results of SCTS and RSHD. The proposed algorithm can track hand or some other nonrigid targets reliably even in clutter environments. Finally, we demonstrate the benefits of our approach in contrast to existing methods.