Research Of Face Recognition Algorithms Based On Local Feature Analysis

The information security science has been attracting much more attention in the last few decades. The identity authentication technology, as an important component part of information security science, is facing the serious challenges, with the fast development of society and the rapid progress of computer technology. Lots of researchers have struggled to find a safer, faster and more convenient identity authentication method, to replace the current traditional methods, which have many shortcomings. In the voice of this demand, the identity authentication methods based on Human Biometric Identification have emerged. Face recognition as a method of authentication is not only safer, cheaper and more convenient, but also confidentiality and non-compulsion, becomes a very important research direction of the identity authentication.This paper researches on Local Feature Analysis (LFA) face recognition method systematically, including the standardization of facial images, the extraction of the local feature and the recognition using local features. And the research on extraction of local feature and recognition using local features are the most important parts of this paper.LFA face recognition takes the Karhunen-Loeve transformation as the foundation, has the characteristics of topology and locality, overcomes fatal weakness of Principal Components Analysis (PCA) face recognition, which is similarly based on Karhunen-Loeve transformation. Thereby it greatly increases the rate of the correct recognition. LFA face recognition is a very important face recognition method based on the statistical methods. Face recognition methods based on the statistical require the face images of high quality, therefore the author carried on a precise standardization or normalization of the face images, before using Local Feature Analysis face recognition.This paper proposes a facial organ extraction algorithm, has extract eyes,nose,mouth and facial contours effectively, can limit the choices of the characteristic points, accelerates the speed of Local Feature Analysis face recognition algorithm greatly.Through the massive experiments proved that, Local Feature Analysis face recognition, using the standardized facial images and the facial organ extraction algorithm, has a high rate of correct identification with higher speed.