基于图像特征方向的PDE去噪方法  

PDE Method Based on Image Feature Direction Applied in Image Smoothing

作　　者：许江[1] 张弓[1] 

机构地区：[1]南京航空航天大学信息科学与技术学院,江苏南京210016

出　　处：《计算机技术与发展》2009年第5期72-74,共3页Computer Technology and Development

基　　金：国防科工委航空科学基金(04D52032)

摘　　要：基于PDE的非线性扩散滤波方法对接近高斯分布的噪声可以取得较好的效果,但对脉冲噪声效果不是很理想。文中应用了基于图像特征方向的正交坐标系,使噪声消除具有了方向性,并提出了新的扩散系数函数,改变了图像边缘阈值,使其能在消除高梯度图像噪声的同时保持了边缘,在一定程度上解决了边缘保持和噪声消除之间的矛盾。Nonlinear diffusion fikering based on partial differential equation（PDE） does well in removing Gaussian noise,but makes no dif- ference to pulse noise. From analyzing the diffusion coefficient of Perona ＆ Malik model, a new diffusion coefficient ruction was put forward by changing the threshold value of edge in diffusion function,and a new orthogonal eoordination based on the direction of image feature was put in use, so it can get rid of the noise of image,and preserves edges better,moreover it overcomes the contradictions between keeping edges and eliminating noise.

关 键 词：偏微分方程 边缘阈值 噪声消除 图像特征方向 

分 类 号：TN911.73[电子电信—通信与信息系统]