This is a new fast and efficient hidden markov model for face recognition.The recognition rate of my project is 96.5% which is better than most of the face recognition techniques.In my project order statistic filter has been used for image preprocessing and singular value decomposition (SVM) has been used for feature extraction.Because of using SVM as feature extraction technique, recognition speed has been faster than any other technique. I have used AT&T database.
Here 7 states (hair,forehead,eyebrow,eye,nose,mouth,chin) have been used in the hidden markov model for recognizing a person rather than 5 states which have been used before which has also helped to improve performance.