A Class of Multi-scale Models for ImageDenoising in Negative Hilbert-Sobolev Spaces

摘要

In this paper, we propose a class of multi-scale variationalrnmodels for image denoising. Our models decompose a given image intorntwo parts: geometric component representing the objects in the imagernand oscillatory component representing the noise or texture. Consideringrndifferent components belong to different scale spaces and oscillatoryrncomponents have small norm in negative Sobolev spaces, we proposernmulti-scale models image denoising in negative Sobolev space. Numericalrnresults show that our models are flexible and efficient for preservingrntexture when denoising image.