摘要
针对纹理图像在去除噪声时,纹理信息容易被磨光,尤其是纹理的线状结构很容易被破坏的问题,提出在小波域改进耦合P-M扩散与相干增强扩散的方法,并用改进的方法对不同的小波子带进行扩散,然后重构,得到去噪图像。数值实验结果表明,本文方法在达到一定降噪效果,保持区域内部较好光滑性的同时,对保持纹理信息、纹理的线状结构及纹理的光滑有很好的效果,说明该方法对纹理图像去噪有较好的效果。
The texture information is easy to be polished when denoising the images, especially the linear structure of the texture is easy to be damaged. A method is put forward that smooth sub-image by using modified method of combi- nation of P-M diffusion and coherence enhancing diffusion in wavelet domain, the different wavelet sub-bands are dif- fused with the improved method and reconstructed, then the denoising image is obtained. Numerical experiments show that the proposed method have better effect on texture image denoising, especially on keeping the texture information, the linear structure and smoothness of the texture.
出处
《激光与红外》
CAS
CSCD
北大核心
2013年第8期943-946,共4页
Laser & Infrared
基金
陕西省自然科学基础研究计划(No.2011JE011)
陕西省教育厅专项科研基金(No.2013JK0602)资助
关键词
小波变换
偏微分方程
图像去噪
纹理
wavelet transform
partial differential equations
image denoising
texture
