An Adaptive Texture-Preserved Image Denoising Model
Chanjuan Liu,Hailin Zou,Caixia Li,Ying Liu,Yilei Wang,Shixiang Jia,Shusen Zhou
DOI: https://doi.org/10.1007/s12652-015-0286-7
IF: 3.662
2015-01-01
Journal of Ambient Intelligence and Humanized Computing
Abstract:Suppressing noise while preserving textures is one of the most important and challenging problems in natural image denoising. Various priors of natural image, such as gradient based prior, nonlocal self-similarity based prior etc., have been widely studied for noise removal. The methods based on these priors may smooth the fine scale image textures and degrade visual quality of the image. To improve image visual quality, an improved texture-preserved total variation (TPTV) image denoising model with an adaptive fidelity item is proposed in this paper. Firstly, we construct an image structure control function (SCF) based on structure tensor to describe the image structure information. Secondly, we combine SCF into a total variation framework for noise removal such that the model can adaptively balance its regular and fidelity item to keep fine scale features while denoising. Finally, extensive experimental evaluations demonstrate that our TPTV model can well preserve the texture appearance in the denoised image and make them more natural. Besides, it overcomes staircase and over-smoothing effects compared with some competing algorithms.