Face Recognition is very booming and very interesting area of security, but due to so many challenges, it is still in the Laboratory experimental work. Face recognition in different illumination is challenging task. Hyperspectral imaging data gives very fine detail information of the face. There is a chance of removing the illumination problem through hyperspectral imaging sensor. In this paper, band wise performance is measured for hyperspectral based face recognition on UWA hyperspectral face database. The proposed method results different accuracy rates at different spectral bands. To reduce the computational burden the proposed method uses PCA algorithm as dimensional reduction technique which also used to select the band randomly and gives an improved accuracy rate of 80.76% at the Band II. The section I gives introduction, section II elaborates research work of the community. The section III and IV discusses mathematical formulation of PCA techniques and proposed methodology respectively in detail. The section V explains experimental work and results however, section VI concludes that through the evaluation of proposed method the recognition rate varies with the selection of different bands.
Volume 11 | 04-Special Issue
Pages: 865-871