Hyperspectral Face Recognition using 3D Discrete Wavelet Transform

Abstract

In this paper a three dimensional discrete wavelet transform (3D-DWT) based feature extraction for the classification of facial hyperspectral imagery is proposed. Most of the relevant work processes 2-D slices of hyperspectral images separately; 3D-DWT has the advantage of extracting the spatial and spectral information simultaneously. Decomposing an image into a set of spatial-spectral components is an important characteristic of 3D-DWT. We propose two methods for 3D-DWT feature extraction, namely, 3D subband energy (3D-SE) and 3D subband overlapping cube (3D-SOC). Extracted feature vector datasets are processed through k-NN classifier and their performance is evaluated under three different testing scenarios. The experimental results revealed that hyperspectral face recognition with proposed 3D-DWT methods substantially outperforms the methods used in spatial-spectral classification reported in the literature.