Image Representation Using Variants of Principal Component Analysis: A Comparative Study

Linear and non-linear data reduction techniques proved their effectiveness in the field of image analysis. Principal Component Analysis (PCA) is a powerful data reduction and data representation technique having its linear and non-linear counterparts. It is a statistical technique used to transform high dimensional data into low dimensional representation without losing much of the information. PCA is a widely-used algorithm in the field of pattern recognition, face recognition, image fusion, data compression, and machine vision. Over the period, many PCA based algorithms have been proposed to effectively extract important features from images and to reconstruct images using minimum feature set. In this paper, we compared four widely used PCA based algorithms in terms of their capability of representing images using reduced feature set.