%0 Journal Article
%T Face Recognition in Compressed Domain by Applying Wavelet Transform and Feature Vector Optimization
%A Menila James
%A S. Arockiasamy
%A P. Ranjit Jeba Thangaiah
%J Journal of Applied Sciences
%D 2013
%I Asian Network for Scientific Information
%X In this study, a novel method for extracting feature vector in the form of entropy points from compressed images was proposed. This efficient approach for performing face recognition systems directly into wavelet based compressed domain which involve wavelet based image compression/decompression for feature extraction, an efficient feature optimization technique and a method for image classification. This is accomplished by stopping the decompression process after entropy decoding and utilizing the entropy points as input to recognition systems. During the experiments, firstly, a standard recognition algorithms used like principal component analysis, independent component analysis and kernel PCA for optimizing feature vector and kd-tree based method for image matching. Secondly, an improved version of canonical correlation analysis method is applied for feature projection. Finally, cascade forward neural network based algorithm for better matching of facial images for image classification. The experimental results proved that the proposed approach is effective in achieving face recognition in compressed domain with additional reduction of computational time and storage requirements.
%K entropy points
%K Face recognition
%K compressed domain
%K tree matching
%U http://docsdrive.com/pdfs/ansinet/jas/2013/451-457.pdf