Face Recognition Algorithm Based On The Video Single Sample Mode

In recent years, because of the advantages of the direct, rapid, non-contact operation and wide application prospect, the face recognition technology has become an important topic in the field of computer image processing. A typical face recognition system mainly includes four parts, such as face detection, image preprocessing, feature extraction and selection, and facial recognition. This paper focuses on the four parts. On the VS2005development platform by calling the OpenCV open-source library for MFC programming, a design has realized the face recognition system in a single sample mode.The face recognition system designed in this paper, is divided into two parts:human face registration system and the face recognition system. Face registration system is mainly used to collect the samples of the different faces, and the feature data is saved in XML file. Face recognition system compares the face image feature from the feature library with the feature of face images in the camera, and determine that whether they belong to one person.Face registration system and face recognition systems need for face detection, image preprocessing and feature extraction and selection. Firstly, the image from camera is converted to grayscale, and AdaBoost algorithm is used to detect the face; If there exists a face, ASM model is used to match the face image in order to get the coordinates of the center of the left eye area and the right eye area, and then the coordinates of the eyes are used to rotate and zoom the face image; Homomorphic filtering algorithm and histogram specification algorithm are used to deal with the light the face image after correction and normalization, in order to cut the influence of uneven illumination; Gabor filter is used to extract the features of face images through the light treatment. Face registered system need to extract the feature and save it in XML format file; Face recognition system compares the face image feature from the feature library with the feature of face images in the camera to determine whether they are from one person according to a certain threshold.The innovation of this paper mainly has two points:first of all, this paper designed and implemented the face recognition system which is based on the single sample mode, that is to say, with only one face image feature of different people, the human face recognition system can achieve higher recognition rate; The second, After using ASM model to corrected and normalized the face image, we can get68key feature points which is including their eyebrows, eyes, nose, mouth, and outline etc. Make full use of the coordinates of the key point’s when extracting the feature of the image. Because the open and close of eyes and mouth easily lead to local texture changes greatly, we respectively extract the Gabor features of the key point of the eyebrows, eyelids, nose in feature extraction.