De-Noising Coloured Images Using Modified Ridgelet Transform

Abstract

The work aims at designing Ridgelet transform  based denoising algorithms using Radon projections. The Ridgelet coefficients are averaged to achieve denoising.  The denoising is performed  using one dimensional  Discrete Wavelet transform (DWT1D) algorithm  for varying levels of White Gaussian Noise (WGN) and Speckle noise for variety of images.  The work also  deals with denoising using cycle spinning techniques with wavelet shrinkage.  The study is carried out using the original image's Mean square error and Peak signal to noise ratio, noisy image and replicated image with and without loop spinning algorithm. The experiments show that  the cycle spinning with wavelet shrinkage yields consistently better results for gray and color images and further points to possibility of scope of research in this area.