Abstract:  An automatic track and recognition system of human faces with five freedom degrees was developed to rapidly and correctly recognize a person．In this system‚the skin-color model combined with the comparability degree and the face figure was adopted to detect and segment a human face‚and the embedded hidden Markov model was ap-plied to the extraction and recognition of face characteristics．The experimental results indicate that the system can not only track‚locate and recognize a human face automatically‚but also confirm an identity on time by comparing the recognized images with those in the face database．The proposed system can be used to reconnoitre criminal cases and monitor the entrances of buildings．

Key words: human face detection, track, recognition, hidden Markov model, image processing