摘要
提出了一种新的图像恢复模型.首先对含噪图像进行各向异性扩散,求得光滑后图像的张量场,然后对求得的张量场和噪声图像进行拟合重构.从而克服了各向异性扩散和结构张量的缺点.数值实验表明,本模型在降低图像噪声的同时,能够更好地保留图像的边缘和纹理.
A novel model of image restoration is proposed.The structure tensor field is calculated after smoothing the noised image by anisotropic diffusion.Then the image is reconstructed by fitting the structure tensor field and the noise image.The proposed method overcomes shortcomings of the classical anisotropic diffusion method.Numerical experiments verify that the proposed method remove noise effectively while preserving structures well.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2011年第6期68-74,共7页
Journal of Xidian University
基金
国家自然科学基金资助项目(60872138)
西安电子科技大学基本科研业务费资助项目(72105370)
关键词
图像去噪
各向异性扩散
结构张量
加性算子分裂
image denoising
anisotropic diffusion
structure tensor
additional operator splitting