Noise Filtering of Images Using Generalized Singular Spectrum Analysis

Murotani,K., Yagawa,G.

Abstract:
There is a noise filtering method for images using Singular Value Decomposition (SVD). This is the method that pixel values of an image are regarded as elements of a matrix, the image is separated into rough parts and detailed parts by SVD of the matrix and the detailed parts of the image are regarded as noise and removed. Generalized Singular Spectrum Analysis (GSSA) is the method that generalizes SVD to treat more generalized data structure than in SVD. In this research, we present noise filtering methods of images using GSSA.