对人脸识别的贝叶斯方法ML中相似度计算公式进行了简化，对数据集的训练和人脸图像的预处理进

行了修改，提出了一种改进的贝叶斯人脸识另1】算法SML。在FERET人脸图像库的子集和南大人脸图像实验库上对

识别算法进行了测试和比较。实验表明，SML算法提高了ML算法的效率，克服了ML算法计算效率不高的缺陷，而

且SML的识别效率明显高于PCA方法。-Bayesian face recognition method on the ML in the similarity formula has been simplified, the data set of training and pre-processing face images were modified, an improved understanding of Bayesian Face Ling 1】 algorithm SML. In the FERET face image database a subset of the Southern adults face image recognition algorithm on the experimental database was tested and compared. Experiments show that, SML algorithm improves the efficiency of ML algorithm, ML algorithm overcomes the shortcomings of high efficiency, and the recognition rate significantly higher than SML PCA method.