Study On Detection And Recognition Methods Of Face Image Including The Electro Magnetic Interference

Because of the extensive application foreground in identity authentication, criminal identity, video meeting, intelligent monitoring and so on, face detection and face recognition has been an important research problem in image treatment and pattern recognition. In the thesis, the methods of face detection and recognition from images including the electromagnetic interference are mainly researched.This dissertation introduces background and present status of face detection and face recognition technology, summarizes the methods of face detection and face recognition, and point out the particularity of face detection and recognition in the environment of electromagnetic interference.Firstly,the theory and the source of electromagnetic interference is analyzed,the kind and the characteristic of Electromagnetic Interference is expounded,the methods of radiate interference and conduct interference is researched.Secondly,the source of image noise, which is interfered by the electromagnetic, is analyzed , and the model of noise is established. Secondly, the basic theory of wavelet transform is researched, and the characteristic of wavelet transform is analyzed. It is researched on wavelets application in the fields of image denosing in electromagnetic interference. the results are analyzed through experiment.The face detection is in progress based on the image which has been denoised. A method for face detection was proposed using both wavelet transform and template matching.First use wavelet transform, then set region value after getting low-frequent coefficient,then come into being some inquire region in two value.Then circumgyrate transform and flex transform for template.Lastly template and inquire region are matched.This method has high speed in complex background, and no influence when face circumgyrate. Experimental results demonstrate its effectiveness.The eigenface method has been improved in the recognition process. Due to the localization that only recognize standard and obverse faces using eigenfaces, anterior dispose module was designed. The face recognition experiment with the ORL face databases shows that algorithm is fast and the recognition inaccuracy of system is low and good effect.