In our recent work, a class of parametric transforms, including family of Haar-like transforms, was introduced and studied in application to image compression. Parametric Haar-like transform (PHT) is a discrete orthogonal transform such that it may be computed with a fast algorithm in a structure similar to that of classical fast Haar transform and such that its matrix contains arbitrary predefined vector in its first row. The aim of this paper is to study a potential use of parametric Haar-like transforms in image denoising. A PHT-based post-processing method is proposed which may improve a denoising method based on fixed transforms. In particular, it is shown that the proposed method may significantly improve the performance of wavelet thresholding based image denoising.
Karen O. Egiazarian, Jaakko Astola