Two-Dimensional Discriminant Supervised Neighborhood Embedding (2D-DSNE) For Face Recognition

DOI Web Site 8 References

Search this article

Abstract

Face recognition has become a very active and attractive research area in recent years due to its wide applications such as using as a password in computer, mobile telephone. In application, a realtime system is a big requirement for several systems, especially in embedded system that microprocessors are not very strong. For example, recognize a person through image from camera in mobile telephone that more and more popular nowadays. Several methods have been developed in recent years, and among various face recognition algorithms, one of the most successful techniques in realtime applications has been the appearance-based method. When using the appearance-based method, an image of size (m×n)-pixels by a vector in an (m×n) dimensional space is presented. In practice, however, these (m×n) dimensional spaces are too large to allow rapid and robust face recognition. A common attempt to solve this problem is through the use of dimensionality reduction techniques. For example, some of the most popular techniques can be used for this purpose including PCA, LDA, etc… However, there are a lot of drawbacks in these methods. We propose a novel algorithm, A Two-Dimensional Discriminant Supervised Neighborhood Embedding, which ensures two purposes: Rapidly process for using in some embedded systems (e.g. in mobile telephone microcontroller, video telephone) and robustly recognize for high recognition rate.

Journal

  • ITE Technical Report

    ITE Technical Report 33.36 (0), 39-44, 2009

    The Institute of Image Information and Television Engineers

References(8)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top