De-noising and recovering images based on Kernel PCA theory

Abstract

Principal Component Analysis (PCA) is a basis transformation to diagonalize an estimate of the covariance matrix of input data and, the new coordinates in the Eigenvector basis are called principal components. Since Kernel PCA is just a PCA in feature space F, the projection of an image in input space can be reconstructed from its principal components in feature space. This enables us to perform several applications concerning de-noising and recovering images. Because of the superiority of Kernel PCA over linear PCA, we also get satisfactory effects of de-noising images using Kernel PCA.