关键词: 人脸识别, Gabor变换, 图像分块, 最大似然准则（ML）, 贝叶斯分类

Abstract: An improved Maximum Likelihood（ML） measure is proposed, which simplifies the similarity computation in the Bayesian algorithm. And then a novel block-based Gabor transformed for Bayesian face recognition is proposed. The original sample images are divided into smaller sub-images, utilizing the convolution of the sub-images and the Gabor filters to extract features, each sub-image is designed as a ML classifier of Bayesian, by use of weighed average similarity to make the final decision. The experiments on FERET face database have shown the effectiveness of the method.

Key words: face recognition, Gabor transform, image-blocked, Maximum Likelihood（ML）, Bayesian classifier