Chapter 9
Evolutionary design for face
recognition
9.1 Introduction . . . . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . .. . 161
9.2 Genetic algorithms . . . . . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . .. . . . . 162
9.2.1 Implementation . . . . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . . 163
9.2.2 Algorithm . . . . . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . .. . . . . . 164
9.3 Representation and discrimination .. . . .. . . . .. . . . . .. . . . . .. . . . .. . . . 165
9.3.1 Whitening and rotation transformation . . . . .. . . . . .. . . . .. 165
9.3.2 Chromosome representation and genetic operators . . . . . 167
9.3.3 The fitness function . . . .. . . . . .. . . . .. . . . . .. . . . . .. . . . .. . . . . 167
9.3.4 The evolutionary pursuit algorithm for face recognition 168
9.1 Introduction
Evolutionary pursuit (EP) [131] is a novel and adaptive representation
method for image encoding and classification. EP seeks to learn an optimal
basis for the dual purpose of data compression and pattern classification.
The challenge for EP is to increase the generalization ability of the learning
machine as a result of seeking the trade-off between minimizing the empirical
risk encountered during training and narrowing the confidence interval for
reducing the guaranteed risk during future testing on unseen images. EP
implements strategies characteristic of genetic algorithms (GAs) for searching
the space of possible solutions to determine the optimal basis. EP starts by
projecting the original data into a lower dimensional whitened image space
obtained from principal component analysis (PCA). Directed but random
rotations of the basis vectors in this space are then searched by GAs where
evolution is driven by a fitness function defined in terms of performance
accuracy which is again termed as empirical risk for class separation.
Face recognition depends heavily on the particular choice of face features
used by the classifier. One usually starts with a given set of features and then
attempts to derive an optimal subset of features leading to high classification
performance with the expectation that a similar performance will be displayed
also in future trials on other datasets. The process of feature selection
involves the derivation of salient features with the twin goals of reducing
161

Get Face Detection and Recognition now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.