The Research Of Face Recognition Technology Based On Local Texture Feature

Face recognition with a wide range of applications is a research focus in computervision and pattern recognition. In this paper, some studies and research have been doneto the face recognition technology. With some current excellent algorithms in facerecognition field, a novel face detection algorithm based on skin color detection andAdaBoost face detection and a face recognition algorithm based on2D Gabor waveletand Local Binary Pattern are proposed.In this paper, Gauss skin color model and ellipse skin color model are improved. Askin color model combined them is used to test color regions of an image, and then thecontours of the regions are found. Then, use the cascade Real AdaBoost classifierconstructed with Haar feature to detect face in color regions; transform gray-scale faceimages detected with2D Gabor transformation and LBP transformation; divide theimages which are generated after transformations into grids; compare the histogram ofeach grid; get the weight of each grid; match faces by weighted K-nearest neighbor.Experiments improves that this method has a higher recognition rate comparing withusing a single transformation and non-weighted method. In order to reduce the numberof histogram comparison, this paper presents a method based on the sequence ofabilities to distinguish faces to exclude faces successively.End of this paper, a camera-based face recognition system is implemented. Thissystem is able to capture images, detect faces in these images, and recognize them. It isproved that this system achieves a high recognition rate.