Dictionary-based anisotropic diffusion for noise reduction

Sung In Cho, Suk-Ju Kang, Hi-Seok Kim, Young Hwan Kim

Abstract: Highlights•Pre-trained dictionary-based anisotropic diffusion is proposed for noise reduction.•Multiscale region analysis is proposed for an efficient patch classification.•The anisotropic diffusion is transformed into a single-pass adaptive smoothing.•In the results, the proposed method showed an outstanding denoising performance.