A Gabor Pseudo Fisherface Based Face Recognition Algorithm for LSI Implementation

Abstract

A face recognition algorithm for LSI implementation suitable for embedded applications is presented. Although many algorithms to recognize a face [1,2,6]recently suggested have a relatively stable performance under the variety of environmental disturbances, the problem still lies on computational cost as well as memory usage. In this paper, we will propose an algorithm based on a "Pseudo Fisherface Matrix", which is derived from generic datasets and down-sampled Gabor features, which reduces these costs. We will compare performance of the proposed method with a traditional one based on Eigenface through FERET database. An experimental implementation demonstrated the proposed algorithm draws significant performance rations.

VLSI processor with a configurable processing element array for balanced feature extraction in high-resolution images

“…In addition, the complex algorithms developed without considering their hardware realization are difficult to be directly implemented into highly efficient very large scale integration (VLSI) processors. In all VLSI processors reported using such algorithms, the original feature extraction process is either modified to fit the hardware implementation [5][6][7][8] or carried out only in some small regions of interest (ROIs). [9][10][11][12] Therefore, much effort is needed to maintain the image representing performance of the simplified algorithms.…”