Eigenface Optimization Research for Variable Illumination Conditions
不同光照条件下特征脸方法的改进研究

A pre-processing method for face images is introduced firstly and the principal components of the eigenfaces are researched for variable illumination conditions. The source and the realization of the eigenface method are introduced and the pre-processing step of the zero mean of every face image is added. Based on the feature database and the probe database from FERET face images, the first three principal components are analysed and the rules of their influences by variable illumination conditions are concluded. The optimal method of choosing principal components is presented when influenced by illumination conditions severely.