Image Denoising Algorithm Using Neighbourhood Characteristics and Cycle Spinning

A new adaptive shrinkage denoising approach based on contourlet transform is presented. Classical contourlet-based denoising method processes the contourlet coefficients with a fixed threshold in each sub band, without considering the neighbourhood characteristics of the coefficients, and introduces artifacts due to the lack of translation invariance of the contourlet transform. We propose an improved shrinkage threshold in contourlet domain which makes good use of the clustering property of contourlet coefficients, and use cycle spinning to compensate for the lack of translation invariance of the contourlet transform. Thus, the new denoising algorithm proposed achieves better tradeoff between details retain and noises removal, and effectively reduces the artifacts. Experiments on test images show that the proposed method outperforms the classical contourlet-based denoising method, in terms of both PSNR values and visual quality.