Image Set Representation with L1-Norm Optimal Mean Robust Principal Component Analysis

Many problems in computer vision area can be formulated as image set representation and classification. One main challenge is that image set data usually contains various kinds of noises and outliers which usually make the recognition/learning tasks of image set more challengeable. In this paper, we propose a new L-1 norm optimal Mean Principal Component Analysis (L1-MPCA) to learn an optimal low-rank representation for image set. Comparing with original observed image set, L1-MPCA based low-rank representation is generally noiseless and thus can encourage more robust learning process. An effective update algorithm has been proposed to solve the proposed L1-MPCA model. Experimental results on several datasets demonstrate the effectiveness and robustness of the proposed L1-MPCA method.