Home
This Title All WIREs
WIREs RSS Feed
How to cite this WIREs title:
WIREs Comp Stat

Biometric face recognition: from classical statistics to future challenges

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

Face recognition involves at least three major concepts from statistics: dimension reduction, feature extraction, and prediction. A selective review of algorithms, from seminal to state‐of‐the‐art, explores how these concepts persist as organizing principles in the field. Algorithms based directly upon classical statistical techniques include linear methods like principal component analysis and linear discriminant analysis. Nonlinear manifold methods, such as Laplacianfaces and Stiefel quotients, offer considerable performance improvements. Other noteworthy ideas include three‐dimensional morphable models, methods using local regions and/or alternative feature spaces (e.g., elastic bunch graph matching and local binary patterns) and sparse representation approaches. Opportunities for innovative statistical and collaborative research in face recognition are expanding in tandem with the growing complexity and diversity of applications. WIREs Comput Stat 2013, 5:288–308. doi: 10.1002/wics.1262 This article is categorized under: Applications of Computational Statistics > Defense and National Security Software for Computational Statistics > Artificial Intelligence and Expert Systems Applications of Computational Statistics > Signal and Image Processing and Coding
An example of Eigenfaces. The top left image is the original face and the remaining ones are the first seven eigenvectors. Because eigenvector elements correspond to pixels, they may be visualized as images and the patterns can suggest what is being encoded. For example, the first eigenvector is primarily encoding lighting variation from left to right.
[ Normal View | Magnified View ]
An example of LBP calculation. (Reprinted with permission from Ref . Copyright 2008 IEEE)
[ Normal View | Magnified View ]
Key aspects of the local region PCA algorithm. Self‐quotient normalization is illustrated in the top left. The local regions are shown on the right. A separate PCA subspace is developed for each local region and for the overall face. The first few eigenvectors for the whole image as well as three of the local regions are shown in the bottom left.
[ Normal View | Magnified View ]
Key aspects of the EBGM algorithm.
[ Normal View | Magnified View ]
Illustration of fitting a three‐dimensional morphable model to a new face image. The bottom row shows renderings of the model from different views. (Reprinted with permission from Ref . Copyright 2003 IEEE)
[ Normal View | Magnified View ]
Illustration of the relighting process that creates nine alternate images using a generic lighting model and a single image.
[ Normal View | Magnified View ]
An example of Fisherfaces discriminants. As with eigenfaces, the Fisherfaces basis vectors may be visualized as images and doing so highlights the patterns being selected in order to maintain separation between subjects.
[ Normal View | Magnified View ]

Browse by Topic

Applications of Computational Statistics > Signal and Image Processing and Coding
Software for Computational Statistics > Artificial Intelligence and Expert Systems
Applications of Computational Statistics > Defense and National Security

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts