关键词: 二维化，主成分分析，线性判别分析，模糊集，隶属度

Abstract: In this paper, a novel method based on fuzzy 2DPLA was proposed which embeds the degree matrix of membership based on the fuzzy set into 2DPCA and 2DPCA methods. This method first calculated the degree matrix of membership using the fuzzy KNN method. In the sequcl,the degree matrix of member was embedded into 2DPCA and 2DLDA for reducing the dimension from horizontal and vertical two directions of image matrix, respectively. Last but more importantly, we used the F-norm classification measure based on the matrix instead of the traditional 2-norm measure based on the vector. In experimental phase, Yale Face Database B, ORL and FERET Face Database were used to test the new method. Experimental results demonstrate that our method has better robustness and higher recognition rate than the state-of-art methods.

Key words: Two-dimensional, Principal component analysis, Linear discriminant analysis, Fuzzy set, Degree of membership