Research And Application Of Face Detection And Recognition Algorithm Based On NIR Image

Because of the advantages of immediate, convenient, friendly, easy to accept, etc.,face recognition technology has always been the research hotspot in the fields ofcomputer vision, pattern recognition and biological recognition, and has greatapplication prospect in the aspects of personal identification, information security andsocial security, etc. In face recognition, the influence of illumination change is the keyfactor of rejection of the recognition performance improvement. At present, one of theeffective solutions of the light problem is to use the active near-infrared (NIR) light forface recognition, which can provide a light homogeneous NIR image for face detectionand recognition, thus greatly reducing the impact of the visible light changes.In this paper, the researching and improvement of the algorithms is conduct aroundthree aspects: face detection, feature extraction and feature classification. In the part offace detection, we choose the Active Shape Model (ASM) based face detection method,and do some improvement for the traditional algorithm. Using the rough positiondetected by the Haar feature based classifier as the shape model’s initialization position,which can reduce the iteration times and increase the speed of face detection to a certainextent, and it can be proved that the ASM method has good robustness for the changesof face expression and posture. For feature extraction, since Gabor Feature has verygood ability of describing the signal characteristics in time and frequency domain, wechoose Gabor Feature instead of the gray pixels as the recognition feature. For the partof feature classification, after comparing the merit and demerit of PCA and LDAalgorithm, we present a novel method combining PCA and LDA, which can make fulluse of the PCA advantage of reduction for the high dimensional Gabor Feature, and cansolve the big matrix and within-class scatter matrix singular problems which is causedby LDA. And PCA+LDA method is proved that it can improve the performance of facerecognition.On the basis of the above study, we build a NIR-Image based face recognitionsystem on Microsoft Visual Studio, which can process static images and real-time video.At last we conduct an experiment to evaluate the performance of the system in threedifferent light conditions. The experimental results prove that the system have goodrobustness to different light conditions.