Singular spectrum analysis for image processing

A digital image is generally encoded as a matrix of grey level or colour values. In the case of grey level images, each element of the matrix is a grey level, in the case of colour images each element of the grid is a triplet of values for red, green and blue components. The two main sources of errors in image processing are categorized as blur and noise. Blur is intrinsic to image acquisition systems as digital images have a finite number of samples. Noise is another source of error; as indicated below there are different types of noise. The search for efficient denoising methods is still a valid challenge in image processing [1]. In this paper we use the celebrated ‘Lena’ image to illustrate the performance of SSA for the problem of image denoising. We also introduce SSArelated distances between images and argue that these distances can be much more informative than the currently used L2-based distances.