Fast blockwise SURE shrinkage for image denoising

In this letter, we investigate the shrinkage problem for the non-local means (NLM) image denoising. In particular, we derive the closed-form of the optimal blockwise shrinkage for NLM that minimizes the Stein’s unbiased risk estimator (SURE). We also propose a constant complexity algorithm allowing fast blockwise shrinkage. Simulation results show that the proposed blockwise shrinkage method improves NLM performance in attaining higher peak signal noise ratio (PSNR) and structural similarity index (SSIM), and makes NLM more robust against parameter changes. Similar ideas can be applicable to other patchwise image denoising techniques such as [1] and [2]. Index Terms Image Denoising, Non-Local Means, Adaptive Algorithm, Shrinkage Estimator, Stein’s Unbiased Risk Estimator Brian Tracey and Joseph P. Noonan are with the department of electrical and computer engineering, Tufts university, 161 College Ave, Medford, MA 02155; e-mail: [email protected]. Premkumar Natarajan is with the Raytheon BBN technologies, 10 Moulton St., Cambridge, MA 02138. Yue Wu was with the department of electrical and computer engineering at Tufts university, but is now with the Raytheon BBN technologies. ar X iv :1 30 5. 42 98 v1 [ cs .C V ] 1 8 M ay 2 01 3