Multiplicative noise removal using self-organising maps
Michel Haritopoulos, Hujun Yin, Nigel M Allinson · Research Explorer (The University of Manchester) · 2001
This paper approaches the problem of image denoising from an Independent Component Analysis (ICA) perspective. Considering that the pixels intensity constituting the crude images represents the useful signal corrupted with noise, we show that, a nonlinear ICA-based approach can provide a satisfactory solution to the Non-Linear Blind Source Separation problem (NLBSS). Self-Organizing Maps (SOMs) are well suited for performing this task, due to their nonlinear mapping property. Separation results obtained from test images demonstrate the feasibility of the proposed method.