| The application of face recognition in network monitor of digital video has become a mostly research content of computer science and technique. At present, face recognition is an import problem of pattern recognition domain and an active research direction present. It is used widely in many domains such as network monitor system of digital video, information security, crack criminal case, bank security, door-guard systems.Facial feature extraction is import part in face recognition. Definition of facial feature extraction affects direct rate of face recognition. It is difficult to obtain satisfying recognition rate in different moving angle, illumination and expression. So we use it difficultly into engineering. The paper started art angle, proposed based on art angle adopting template matching method. Our result of recognition reach 86% on part of 640*480,1350 people, 24500 images having complex background. We proposed multi-objects optimization algorithms of facial feature extraction, According to our test results ,we are satisfied with taking conclusion of improving 2%~5% accuracy.On design face recognition network monitor of digital video, we adopt embedded WEB server. The system according to real-time of video data, synchronization of video terminal, agility of expansion, celerity, cellular and stability, combine to operating speed and accuracy of DSP, use the face recognition and optimization method proposed, adopt DSP develop facility series of C6X .We can get a capability predominant network monitor system with face recognition after adding into it.
|
PDF Full Text Request