Pca-based Face Recognition Algorithm Research

Automated face recognition on computer is an extremely challegeable research project, which assembles the knowledge and technology of various scientific subjects, such as signal processing, intellectual control, pattern recognition and machine vision, etc. How to implement rapid and accurate face recognition with computer has been a hotspot as well as a difficulty in the fields of image processing and pattern recognition all along.This thesis carries out a series of researches concerning the existing problems in the process of face recognition; the main contributions are as follows:1) Presents a method of adjusting face position on the basis of single face eye location in grayscale images. It is designed according to the features of human eyes' grayscale variations, geometric structures and also the axial symmetry of two eyes. Experimental results have proved that for single face images with both eyes visible, this method can realize face position adjustment fase and efficiently.2) Proposes a new face recognition algorithm based on modular principal component analysis (modular PCA) together with an improved mesurement of the distance of subfaces. Compared with the traditional PCA method, it can obtain a higher recognition rate when the illumination...