Concurrent Self-Organizing Maps for Multispectral Facial Image Recognition

This paper is dedicated to multispectral facial recognition, based on the model of concurrent self-organizing maps (CSOM), previously proposed by first author. The first approach of this paper is to apply CSOM classifier for color face recognition. Main variant of this approach has the follwing processing stages: (a) color conversion from the 3D RGB space into an optimum 2D selected color feature space; (b) principal component analysis (PCA) for each resulted color component; (c) feature fusion; (d) CSOM/SOM classification. The proposed system is experimented using the ESSEX database of color facial images; it contains 151 subjects, where each is represented by 20 pictures of 200 times 180 pixels. The obvious advantage of CSOM over SOM is proved. The second approach of this paper is the implementation of a real time CSOM face recognition system using the decision fusion that combines the recognition scores generated from visual channels (R, G, B, and Y classifiers) with the thermal infrared classifier. As a source of color and infrared images, we used our VICE ACE database of 38 subjects. Any picture has 160 - 120 pixels; for each subject there are pictures corresponding to various face expressions and illuminations, in the visual and infrared spectrum. The spectral sensitivity of infrared images corresponds to the longwave range of 7.5-13 mum. The very good experimental results are given, proving nearly invariance to illumination conditions