Stationary wavelet transform based on image rotation and its application to image denoising

There are two different problems in image denoising using wavelet transform most subsampled filterbanks lack the proper of translation invariance, and the textures and edges of denoising image become blurry. To overcome the shortcomings, a novel denoising algorithm using stationary wavelet transform based on image rotation (RSWT) is presented. Firstly the RSWT is performed by adopting stationary wavelet transform on 2-D rotated image to success in obtaining image directional singularities information. Then, we employ bivariate shrinkage scheme to the RSWT and achieve an efficient image denoising approach. Experiment shows the proposed denoising algorithm removes most of the noise and produce superior visual quality than other methods (DWT, SWT), especially for images that have abundant textures and edges.