摘要： 提出了一种综合利用整体和局部信息进行人脸识别的新方法。在对整幅人脸图像进行PCA分析的基础上辅以了局部区域的PCA+LDA分析。在ORL的400幅人脸库上对此方法进行了验证，结果证明此方法是有效可行的，最优的识别率达到了97%，比仅利用完整图像的人脸识别方法有了不同程度的提高。

关键词: 人脸识别, 主分量分析, 线性鉴别分析

Abstract: This paper describes a face recognition algorithm based on both global and local information. The algorithm proposed here combines global and local information together. The paper applies PCA to the whole face image while applying PCA combining LDA to some local regions of the image. The algorithm is tested on ORL face database, which contains 400 face images. It achieves a recognition rate of 97%, and demonstrates an improvement compared with the standard algorithm based only on global information.

中图分类号: 

• TP391.41