SVD aided eigenvector decomposition to compute PCA and it's application in image denoising

Abstract

Principal Component analysis (PCA) is a powerful nonparametric tool in modern data analysis which is widely used in diverse fields from neuroscience to image processing. PCA can be calculated in two different ways: decomposition of eigenvectors and singular value decomposition (SVD). In this paper, we proposed a new method of PCA calculation using both SVD and decomposition of eigenvectors. We presented how the proposed method of calculation of PCA improve the performance of PCA in image denoising. We also showed that the proposed method produced better results than the state-of-the-art image denoising algorithms in terms of PSNR, SSIM and visual quality.