Face recognition using orthogonal transform coefficients of hyperspectral face images

Hyperspectral images are used widely in biometric research because of the immense amount of unique data they generate which has proved to be helpful in solving the drawbacks of existing biometric systems. The main focus of the this research is to use hyperspectral face images having 33 bands for face recognition where orthogonal transform coefficients for feature vector generation along row, column and diagonal values and compare the results obtained for single instance and multi-instance analysis with existing systems. With the use of hyperspectral face images, the accuracy rate was found to be improved. However the main drawback of these Hyperspectral images is that they generate large amount of redundant data and hence row, column and diagonal mean values are computed instead of full image so as reduce the memory and time constraints. Orthogonal transforms such as Fast Walsh Transform, Discrete Cosine Transform, Kekre Transform, Fast Hartley Transform and Haar Transform are used for texture feature extraction and coefficients are computed for the row, column and diagonal mean vectors. The extracted feature vectors are put for intra class and inter class testing using Euclidian Distance measure. The performance of the system was measured and compared with existing system. The proposed system has given up to 81.9% results for TAR-TRR for F+L+R multi instance implementation.