Image Denoising Using Adaptive PCA and SVD

The effectiveness of an image denoising algorithm depends upon how the signal is represented in it. A lot of work has been done in the field of image denoising already, but there is a lot of scope for further investigation as well. In this paper, a simple, efficient Patch based and Block based image denoising algorithms, where the noisy image patches are represented using Principal Components and Singular Values is presented. From the conventional Principal Component Analysis (PCA) based denoising algorithm two improved versions of denoising algorithm were developed using patch based and block based Singular Value Decomposition (SVD). These techniques were found to work excellently on images affected by different kinds of noises. A comparison of the three methods using a quantitative analysis in terms of PSNR and RMSE is done.