期刊名称：Indian Journal of Computer Science and Engineering
印刷版ISSN：2231-3850
电子版ISSN：0976-5166
出版年度：2020
卷号：11
期号：1
页码：66-75
DOI：10.21817/indjcse/2020/v11i1/201101057
出版社：Engg Journals Publications
摘要：Non-Local Means algorithm (NLM) is a prominent image denoising algorithm. One of the majorlimitations of NLM algorithm and its variants is the time requirement. In this era of high performancecomputing, an efficient alternative to reduce the time complexity of any algorithm is its parallelization. Inthis paper, a parallelized version of basic NLM algorithm using CUDA architecture is proposed. Thealgorithm is developed on NVIDIA GeForce 940M GPU which follows Maxwell architecture with 3 SMsand 384 CUDA cores. Experiments are carried out using selected set of natural and medical images ofvarious sizes. Our proposed parallelized version of NLM algorithm reduces the time requirementapproximately by 50% in comparison to its basic version and also achieves comparable denoisingperformance in terms of PSNR, SSIM and FSIM evaluation metrics. The proposal is a model which canbe customized for newer GPU architectures.
关键词：Image Denoising; Non Local Means Filtering; GPGPU; NVIDIA; CUDA
