改进的各向异性复扩散模型的医学图像去噪方法  

Improved Anisotropic Complex Diffusion Model for Medical Image Denoising

作　　者：张美玉[1] 张素琼[1] 秦绪佳[1] 徐晓刚[2,3] 

机构地区：[1]浙江工业大学计算机科学与技术学院,浙江杭州310032 [2]大连舰艇学院装备系统与自动化系,辽宁大连116018 [3]浙江大学CAD&CG国家重点实验室,浙江杭州310027

出　　处：《小型微型计算机系统》2010年第5期969-973,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60673063)资助;国家科技支撑计划项目(2007BAH11B02)资助;浙江省自然科(Y1080436)资助;浙江省科技计划项目(2009C32007;2009C31106)资助

摘　　要：对医学图像进行有效的去噪并保持边缘信息,有利于图像的后续处理.本文分析P-M模型和Gilboa的复扩散模型以及它们的不足,提出一种改进的各向异性复扩散模型.该方法先用中值滤波对图像进行预处理,去除梯度值大的噪声点,然后用图像的虚部求扩散系数,以此引导扩散模型中的边缘检测函数,再进行八邻域像素的扩散过程.实验表明,该方法能达到较理想的去噪和保持边缘的效果,而且减少了迭代次数,缩短了计算时间.For the purpose of further processing,it is necessary to remove noisy while keeping the edge＇s information in medical processing. In this paper,the P-M model and Gilboa＇s complex diffusion model as well as their disadvantages are discussed. An improved anisotropic complex diffusion model for medical image denoising is proposed. The method applies median filter to pre-processing the image for removing the pixels with high gradient firstly. Then use the image＇s imaginary part for diffusion coefficient solving,so as to adjust edge detection function of the diffusion model. Finally,diffuse the image with eight-neighborhood pixels. Experiments show that the method can remove noises and keep the edge of the image. Also,it can reduce the iterations and shorten the computing time.

关 键 词：医学图像去噪 偏微分方程 P-M模型 各向异性复扩散 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]