Study On Face Detection And Recognition

Personal identification system based on Biometrics, because of using the proper living creature characteristic of human body, is the totally brand-new technique different from traditional method. Because it has the better safety, dependable with the usefulness, more and more people thoughtful of, and the beginning enter our social each realm, greet the modern year's challenge.In the daily life, because people identify surroundings use at most of is a person's face, it is a body for most easily drive accepting that person's face identify method. The person's face is a human sense of vision inside the forest the widespread mode. Function for exchanges for sense of vision information for reflecting in the person and person, have getting the importance of person face with the meaning. The face recognition technique try to make the computer have the identifying of person ability, because its extensive and applied realm, face recognition technique got the extensive concern with study in near three decades. Along with the network technique and the table video extensive adoption, video camera become the standard of the personal computer equipment, at the same time electronic commerce requesting body verify bring up newly, person's face identification becoming most having the potential one of the verification methods.Learned essays, research papers the domestic and international in recent years concerning person's face detection and recognition, some theories problem to computer identify technique to person's face is analyzed. Aim at to establish the personal identification system of two main technique, face detection and face recognition to proceeds the in-depth researched, and bring upped a method for used for frontal face detection and face recognition in video images. Experiments prove the face detection and recognition method by this dissertation bring up is reasonable, and have the certain theories value with the practical value. The textual main research work primarily includes below a few aspect:1.Pass to research domestic and internationally at face detection aspect the analysis of the result, discover to now all establish on the machine study on the successful method fundamental of face detection. The Statistics learning theories (STL) that statistics the research is in recent years to get very the tollgates note the small sample circumstance the machine learning theories. It established a complete theories, and adopt to differ from the tradition of according to experience risk minimum (ERM) theories method, but adopt anew according to construction risk minimum (SRM), combine the proof according to the minimum of construction risk consequently have the good generalization ability. Support vector machines (SVM) is a new machine learning method that is established on the Statistics Learning Theories. This dissertation bring upped a kind of adoption DCT transformation coefficients as the classifier input vector on SVM deeply researched, and make use of the improvement's SMO to learn the classifier and establish a face detection method that base on the SVM. This method improves learning and detection efficiency. Examination in some image databases prove this method have very high detection rate, and have the very high theories value and practical value.2. Aim at SVM method demand towards whole image to detection, this dissertation adopting background separate technique with skin color detection technique combine together to get the area which may be have person's face, then adoption SVM method which have very high detection rate to proceed detection. Because of the area is small, this kind of method namely guaranteed the detection rate at the same time detection speed that near to the real time. Because the face detected images are used for face recognition, therefore acquire the quality images relate to the accuracy that recognition. This dissertation grace used a method for evaluation the detected face images, and export the good quality image.3. On the method that right and every kind o...