Two-Dimensional Random Subspace Analysis for face recognition

Abstract

In this paper, we proposed a novel technique for face recognition using Two-Dimensional Random Subspace Analysis (2DRSA), based on the Two-Dimensional Principal Component Analysis (2DPCA) technique and Random Subspace Method (RSM). In conventional 2DPCA, the image covariance matrix is directly calculated via 2D images in matrix form, by concept of the covariance of a random variable. However, 2DPCA reduces the dimension of the original image matrix in only one directions, normally in the row direction. Thus, it needs many more coefficients for image representation than PCA. For solving this problem, many methods were proposed by considering both the row and column directions. We develop another technique to reduce the dimension of 2DPCA feature matrix in column direction by using of randomization in selecting the rows of feature matrix. And the ensemble classification method is used to classify these feature subspaces. Experimental results on Yale, ORL and AR face databases show the improvement of our proposed techniques over the conventional 2DPCA.