Research And Application Of Face Recognition Algorithm

With the development of science and technology, in particular, the rapid increase in performance of computer software and hardware. As a new technology closely related,biometric technology has made tremendous attention and development. How to quickly and efficiently identify an individual’s identity information has become an important topic of researching. Face recognition is a friendly and accurate identification technology, thus becoming one of the hot research projects. Simultaneously people had to face the sharp accumulation of high-dimensional data such as digital images, document data and financial time series, etc. However, limitations of the computing and storage technology and cost, the dimensionality reduction have become a necessary means of dealing with high-dimensional data. The goal of dimensionality reduction is to map the information in high-dimensional space into a lower one, but its certain properties, especially, geometric and topological properties are preserved. The typical algorithms for learning include Principal Component Analysis(PCA), Linear Discriminant Analysis(LDA), Tensor Subspace Analysis(TSA) and Locality Preserving Projection(LPP). Advanced in biometric technology and dimensionality reduction algorithms, face recognition techniques based on dimensionality reduction methods have been developed rapidly in the past few decades.This paper focuses on face recognition algorithm and do some research. The main contents are as follows:Some classical algorithms are made a brief introduction. These methods can’t make full use of abundant and available labeled examples. Throughout this paper, there is considerable interest in the available labeled examples. Proposed a simple, efficient supervised algorithm called multilevel face recognition algorithm(MFRA) which can take full advantage of class labels or other prior information and achieve better recognition results(The effectiveness of the algorithm can be proved by numerical experiments). In this algorithm, the recognition processes are made up of multiple parts. Each part is relatively independent; however, they are interrelated at the same time.This paper presents the other algorithm called New Tensor Subspace Analysis(NTSA). As can be seen from the name of this algorithm, it is an improvement algorithm based on TSA. By means of some ideas in the semi-supervised face recognition algorithm, and make improvements to the original algorithm model.