Image denoising

Abstract

Image Denoising has been a topic of research interest since as early as 17th century. Various techniques of denoising a noisy image have evolved. In fact, the endeavor towards better Image denoising techniques continues to this day (21st century). The thesis attempts to draw attention towards the Independent Component Analysis, in particular and its application in Image denoising. The author proposes Recursive ICA for image denoising. The method helps in determining whether a feedback would improve the efficiency of the system in extracting the independent components off the observed data or not. It is eventually observed that Recursive ICA proves to be of intermediate efficiency between that of ICA and Orthogonal ICA. In fact, the performance of the former approaches the performance of Orthogonal ICA very closely.