Image Processing Algorithm Of Face Recognition

With the rapid development of computer technology, the accuracy, safety andpracticality of personal identification are highly demanded in modern society.Traditional personal identification methods, such as password, IC card, and ATM cardetc., can't satisfy this kind of requirements already. So, biometric identificationtechnology comes into being.Face recognition technology, as the main method of biometric identificationtechnology, is the point of research and application. This paper mainly researched thestatic face image for identification, involved preprocessing, identification methodsbased on principal component analysis (PCA) and two-dimensional principalcomponent analysis (2DPCA), and methods based on infrared and visible image.After summarized the achievement of others, the author has done some work asfollowing:1. On preprocessing stage, for the difficulties that locating the eyes, an algorithmbased on the integral projection function is presented, and tested on the ORL facedatabase. The result is proved this algorithm has high accuracy on locating the eyes.2. On feature extraction stage, an improved algorithm is presented, after detailanalysis the traditional algorithm of PCA and 2DPCA, and tested on the ORL facedatabase. The result shows that the improved algorithm has higher feature extraction speed and recognition accuracy than both the traditional PCA and 2DPCA.3. Aim at the respective advantages of the infrared and visible image in facerecognition, a fusion algorithm based on wavelet transform is presented to fuse theinfrared and visible image. And tested on the Equinox face database, obtain satisfyingresult.