彩色图像乘性噪声去除的高阶变分模型  

High-Order Variational Model for Color Image Multiplicative Noise Removal

作　　者：王加忠 潘振宽 魏伟波 徐娟 WANG Jia-zhong;PAN Zhen-kuan;WEI Wei-bo;XU Juan(College of Computer Science and Technology,Qingdao University,Qingdao Shandong 266071,China)

机构地区：[1]青岛大学计算机科学技术学院,山东青岛266071

出　　处：《计算机仿真》2020年第2期443-448,485,共7页Computer Simulation

基　　金：国家自然科学基金资助项目(61772294);“十二五”国家科技支撑计划项目(2014BAG03B05)。

摘　　要：传统的变分去噪模型中,MTV模型去噪后的图像可以较好的保持图像的边缘,但会有阶梯效应。高阶TC模型可以防止阶梯效应,但是边缘保持不好。采用耦合的MTV模型和高阶TC模型相结合的方法,构造出新的混合模型,并推广到彩色图像乘性噪声去除的高阶变分模型。为提高新模型的计算效率,引入辅助变量和拉格朗日乘子设计了相应的增广拉格朗日算法。实验结果表明,新模型在处理彩色图像时能有效地避免阶梯效应,同时保持图像的边缘和细节。与实验中的传统模型相比,新模型的峰值信噪比和结构相似性指数均有提升。In the traditional variational denoising model,the image denoised with the MTV model can better preserve the edge of the image,but it can lead to a staircase effect.The high-order TC model prevents the stair effect,but the edge retention is not good.A new variational model was constructed by combining the coupled MTV model and the high-order TC model,and it was extended to the high-order variational model of color image multiplicative noise removal.In order to improve the computational efficiency of the new model,the corresponding augmented Lagrangian algorithm was designed with introducing auxiliary variables and Lagrangian multipliers.The experimental results show that the new model can effectively avoid the staircase effect when processing color images,and keep the edges and details of the image.Compared with the traditional model in the experiment,the peak signal-to-noise ratio and structural similarity index of the new model are improved.

关 键 词：图像去噪 阶梯效应 变分模型 拉格朗日乘子 增广拉格朗日算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]