Image Deblurring by Fast Non-local Means and Bregmanrized Operator Splitting
Zhiming Wang, Hong Bao, Zhang Li · International Conference on Electric Information and Control Engineering · 2012
An image deblurring algorithm based on non-local means denoising and Bregman iteration algorithm is proposed. Non-local total variation (NLTV) regularization term in Bregmanrized operator splitting (BOS) algorithm was replaced with l2 non-local norms, and a iterative three step deblurring scheme with fast non-local denoising algorithm was given. Based on the symmetry of weighted Euclidean distance, the most complex distance computation in non-local means was halved for every pair of pixels. Experimental results show that proposed algorithm obtained better results on ten common test images than NLTV regularization, both in deblurring performance (PSNR and MSSIM) and processing speed.