Design and implementation of the face tracking algorithm based on spatio-temporal context learning

Visual tracking via dense Spatio-temporal Context learning algorithm (STC) is a rapid, accurate and efficient visual tracking algorithm, but it's first frame for tracking target is predetermined, so it can not satisfy the actual application well. In this paper, firstly, fast and accurately detecting and locating the face which should be tracked in the first video frame based on face detection method via skin color detection and visual attention theory, and then, using it as the first input frame for STC tracking algorithm for a scale-adaptive tracking the face target. Experiments show that the algorithm is fast, accurate, efficient, and quite robustness of light. It can meet the actual application.