Research Of Face Recognition Algorithm

Face Recognition is one of the most outstanding ability in human vision, with other biometric identification methods and Human perception and computer interaction field has a close contact. Face recognition technology as memory is let the machine has a fan, distinguish the way of others, it is a combination of image processing technology and Feature extraction, etc. in a variety of subject areas. Along with the social for rapid and accurate about automatic authentication actual demand is increasingly urgent, Face Recognition has become a pattern recognition and computer vision field research hot spot.The thesis emphasis on the image processing technology, face recognition based on the geometric features and Hough transform etc. First of all, image processing technology is studied mainly in several image processing methods, the method of according to the actual needs about research on the gray level transformation, filtering processing, image normalization(angle of rotation, the scale of the image cropping, image normalization), in order to extract the target image can meet the requirements of the system; Second is studied based on the geometric feature of face image feature location and face image feature extraction algorithm, to extract the characteristics of the eyes, mouth, nose, through the characteristics of facial image positioning, in the face image has been positioning on the basis of the senses, extract the left cheek and right cheek and the characteristics of elements of the jaw; in view of the Hough transform in the process of face recognition, image complex degree is high, slow processing speed of, puts forward a improved face recognition algorithm based on Hough transform. Using complexity images quickly locate facial features, the concept of using Hough transform curve transformation such as cheeks into point, by adopting the idea of hierarchical Hough transform to reduce the complexity of the algorithm, and established a set of refining feature vector, to achieve the purpose of rapid and accurate identification of target image. In this paper, based on the above research results, in VC++environment, face recognition system is developed. Experiment and system operation results show that the proposed algorithm than the former algorithm of computing speed, high accuracy, the expected results have been achieved.