Robust face tracking based on bayesian probability

It is a tough task to robust track face in dynamic complex environments with an active camera. A tracking method based on Bayesian probability is proposed to real-time follow-up track moving face in dynamic complex environments. At first frontal face is detected using the Adaboost face detector based on Haar-like features, and is used to initialize the tracking module. Face color feature representation based on histogram is introduced. Face tracking is realized with the enhanced Camshift algorithm implemented in the color probability density distribution image (CPDDI). To consider and suppress disturbance from background, Bayesian probability is used to produce CPDDI. Experimental results show that the presented method is well suitable for tracking face with arbitrary pose in dynamic complex environments with an active camera. The method is computationally efficient and can run in real-time speed completely.