Image set representation and classification with covariate-relation graph

Abstract

Recently, image set representation and classification is an important problem in computer vision and pattern recognition area. It has been widely used in many computer vision applications. In this paper, a new image set representation method, named covariate-relation graph (CRG), has been proposed. CRG aims to represent image set with a graph model. Compared with existing representation methods, CRG is more flexible and intuitive. Based on CRG representation, we further achieve image set classification tasks using Kernel Linear Discriminant Analysis (KLDA) and nearest neighbor classification. Experimental results on several datasets demonstrate the benefit of the proposed CRG representation.