Extracting efficient color features for face recognition using evolutionary computation

This paper presents an evolutionary framework as a novel color feature extraction method for face recognition. First, two new color spaces are defined as linear transformations of the input RGB color space. The first color space is defined by one luminance (L) channel and two chrominance channels (C₁,C₂), and the second color space incorporates one luminance channel (L) and three chrominance channels (C₁,C₂,C₃). Genetic algorithms (GAs), driven by a fitness function that evaluates face recognition accuracy, thus search for the optimal transformations from the RGB color space to the LC₁C₂ and the LC₁C₂C₃ color spaces, respectively. The successful application of the proposed evolutionary framework is demonstrated with the face recognition grand challenge (FRGC) databases and the biometric experimentation environment (BEE) baseline algorithm. In particular, when using an FRGC version 1 dataset containing 366 training images, 152 controlled gallery images, and 608 uncontrolled probe images, the evolutionary framework improves the rank-one face verification rate of the BEE baseline algorithm from 37% to 77%. When applied to an FRGC version 2 dataset consisting of 6,660 training images, 16,028 controlled target images, and 8,014 uncontrolled query images, the evolutionary framework improves the face verification rate (at 0.1% false acceptance rate) of the BEE baseline algorithm from 13% to 37%.