基于改进各向异性的水平集医学图像去噪  

Medical image denoising based on improved anisotropic diffusion and level set

作　　者：霍宇欢 伍岳庆[1] 姚宇[1] 

机构地区：[1]中国科学院成都计算机应用研究所,成都610041 [2]中国科学院大学,北京100049

出　　处：《计算机应用》2015年第A01期270-272,共3页journal of Computer Applications

基　　金：四川省科技支撑计划项目(2012GZ0106)

摘　　要：针对成像复杂、噪声突出的医学图像在去除噪声的同时模糊边缘特征的现象,提出了基于改进的各向异性的水平集去噪模型。在水平集去噪模型的基础上,加入了改进的各向异性扩散因子,其中改进的各向异性扩散因子采用了中值滤波平滑后的梯度模替换原始图像的梯度模,对于医学图像中大量的斑点噪声更加有效,并保留了图像的边缘信息。基于Matlab平台对改进算法进行了验证,实验表明,基于改进各向异性的水平集算法在有效去除噪声的同时,非但没有模糊边缘特征,相反地起到增强边缘信息的效果。改进算法优于各向异性算法和中值滤波等算法,提高了图像的信噪比,降低了图像的均方误差,保留了更多细节信息,使得医学图像更好地用于诊断,以及后续的分割等处理。Based on the phenomenon of complex imaging, prominent noise of medical images blurs edge features while removing the noise, this paper proposed a level set model for image denoising based on modified anisotropic diffusion. Denoising model on the basis of level set, added an improved nisotropic diffusion factor. Anisotropic diffusion factor used gradient mode after median filer smoothing to replace the gradient mode of the original image, which is more effective for lots of speckle noise in medical images and keeps the images edge information. The improved algorithm was verified based on the Matlab system. Experiments show that the algorithm of modified anisotropic diffusion combined with level set method in effectively removing noise, enhances the edge information instead of blurring edge features. The modified algorithm is superior to the anisotropic algorithm and median filtering algorithm, which can improve the SNR （ Signal-to-Noise Ratio） of the images, reduce the mean square error and retain the more detail information, making the medical images better for diagnosis, and subsequent processing such as segmentation.

关 键 词：水平集方法 各向异性算法 医学图像 中值滤波 斑点噪声 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]