سال انتشار: ۱۳۹۰

محل انتشار: هفتمین کنفرانس ماشین بینایی و پردازش تصویر

تعداد صفحات: ۵

نویسنده(ها):

Kamran Binaee – DSP Research Laboratory, Department of Electrical Engineering, University of Guilan, Rasht, Iran
Reza P. R. Hasanzadeh – DSP Research Laboratory, Department of Electrical Engineering, University of Guilan, Rasht, Iran

چکیده:

Conventional Non-Local Means (NLM) as one ofthe most powerful denoising filters especially for reduction ofadditive Gaussian noise is not successful in the case ofUltrasound (US) Images noise suppression. In the presence ofadditive Gaussian noise model, the NLM filter uses Euclideandistance similarity criterion to find similar patches andtherefore it is not appropriate for US images which have noisewith multiplicative and signal dependant nature. The moresuccessful version of NLM filter for US images which isknown as Optimized Bayesian NLM (OBNLM) is developedbased on Pearson Distance similarity criterion to measure andfind the similar patches. In this paper, we tried to improve theperformance of NLM filter using appropriate fuzzy similaritycriteria. The proposed filters are evaluated in objective andsubjective manners with both synthetic phantom and realclinical US images. It is shown that the proposed methodshave better ability for noise reduction comparing with theother state-of-art de-speckling filters.