A Human Face Detection System Based On Skin Color Partition And Facial Features Extraction

With the development of technology, the traditional identity recognition cannot meet the increasing demands. The biologic characteristic based identity recognition such as iris and dactylogram needs very strict information collection conditions, which makes it difficult to be used practically. The verification of biologic character based on the face detection has advantages such as friendly, convenient, and simple hardware, etc., which has attracted more and more attentions these years. As the face detection is not only used for face recognition, but also for the Content Search, Digital Video Processing, therefore, it has research significance as well as practical value.Based on the existing research results, a realization framework of face recognition system is proposed, together the importance of face detection to the identity recognition system. This face detection system has functions of complexion modeling and edge detection, etc. In the system, not only a single image can be detected, but also hundreds of images can also be detected at the same time, which makes it convenient to evaluate the overall effect.After discussing the color space of complexion modeling and the complexion modeling related to the color space YCrCb, this thesis focused on analyzing the space YCrCb and its extended space--NYCr″Cb″. Some classical edge detection algorithms are realized by programming. Based on the simulation results, using Canny operator, a method to extract and validate eyes and mouths is given. This system is developed in VC environment with very compact source file and executive file. Besides the basic function, every important module is designed reasonably, which is easy to extend further.Finally, the problems encountered in the research and their solutions are discussed, and the suggestions for future work is also presented.