基于自适应阈值参数的改进各向异性扩散模型  

An Improved Anisotropic Diffusion Model Based on Adaptive Threshold Parameters

作　　者：陈文斌 王燕玲[1] 王娜 刘祎[1] 张权[1] 邵燕灵[2] CHEN Wen-bin;WANG Yan-ling;WANG Na;LIU Yi;ZHANG Quan;SHAO Yan-ling(Shared Provincial Key Laboratory for Biomedical Imaging and Big Data,North University of China,Taiyuan 030051,China;School of Science,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学生物医学成像与影像大数据山西省重点实验室,山西太原030051 [2]中北大学理学院,山西太原030051

出　　处：《中北大学学报（自然科学版）》2018年第5期602-608,共7页Journal of North University of China(Natural Science Edition)

基　　金：国家重大科学仪器设备开发专项(2014YQ24044508);国家自然科学基金资助项目(61671413);国家重点研发计划(2016YFC0101602);山西省回国留学人员科研资助项目(2016-085);山西省青年基金项目(201601D021080);中北大学校基金(XJJ2016019)

摘　　要：针对传统的各向异性扩散PM模型在对图像去噪时会产生阶梯效应,并且会丢失细节的问题,提出了一种改进的各向异性扩散IPM模型.该模型用梯度模、局部方差以及基于差分曲率的自适应边缘阈值参数构造了新的扩散系数,在边缘、细节、平坦区域进行不同强度的扩散.实验结果表明,IPM模型在去噪的同时不仅能有效保持边缘、细节等信息,还能克服阶梯效应,与其他基于偏微分方程的图像降噪算法相比,信噪比(SNR)、峰值信噪比(PSNR)和平均结构相似度(MSSIM)都得到了提高.Aimed at the drawbacks of traditional anisotropic diffusion PM model,it had a staircase effect on image denoising and resulted in the loss of detail.An improved anisotropic diffusion IPM model was proposed.Gradient magnitudes,local variance and adaptive edge threshold parameter based on the difference curvature constructed a new diffusion coefficient,which diffused differently in edge,fine detail and flat areas.The experimental results show that the proposed IPM method performs efficiently in edge and fine details preservation,avoiding staircase effect while getting rid of noise.Compared with other image de-noising algorithms based on partial differential equations,the PSNR,SNR and MSSIM are improved.

关 键 词：各向异性扩散 自适应边缘阈值参数 局部方差 差分曲率 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]