Based On Two-dimensional Color Images Of The Face Detection, Face Tracking

Human face detection, tracking and recognition are challenging problem in the field of human-computer interaction, which have become an active research topics for many organizations all over the world due to forceful promotion from security field recently.In this paper, we have made a deep study on automatic face detection and tracking in color image sequences. A systematic summary of previous work has been given first. Then this paper presents a scheme of face detection based on skin-color segmentation and facial-features verification,The difficult detection task is divided into three steps: lighting compensation,which can optimize the quality of the image to some extent; skin-color segmentation based on a mixed skin-color model and morphological dilation-erosion operation, which are used for coarse location ; facial-features verification with eyes and mouth features information, which are used for fine face detection. In the case of face tracking, we classified video frame as detection frames and tracking frames. In detection frames, skin color and facial features are all considered, while in tracking frames, only skin color make a sense, and it also adopts the linear motion-prediction to the position of the search region in next of frame. Experiment results show that our algorithm gains certain accuracy and detection rate. It's expected that our algorithm can be used in real time applications in further work.