关键词: 人脸识别，二维线性鉴别分析，二维主成分分析

Abstract: Two-way compression project subspace method combining both the Two-dimension Principle Component Analysis (2DPCA) and the Two-dimension Linear Discriminant Analysis (2DLI)八)was proposed for face recognition.This method first transposes the feature matrix after it performs the 2DPCA and then it performs the 2DLDA. Compared with the (2D) } PCA and (2D)2 LDA, this method makes full use of the advantages of the 2DPCA and 2DLI)、.It not only contains the sample category information, but also eliminates the image matrix correlation of the row and column,so that it effectively extracts the row and column recognition information,and mcanwhile,the recognition feature dimension decreases dramatically. The experiment on the ORL and PERET face databases shows that the recognition rate of this method is better than the existing two-dimension feature extract method without influencing the recognition speed.

Key words: Face recognition,Two dimension linear discriminant analysis,Two principle component analysis