摘要
为保证对含有较多纹理信息的图像有好的恢复性能,提出了一种具有非凸非光滑组合正则的图像恢复方法。一方面,利用一阶非凸非光滑正则在恢复图像的同时保护图像纹理信息;另一方面,采用二阶非凸非光滑正则降低一阶非凸非光滑正则引起的分层效应。交替方向法等用来克服该组合正则项非凸非光滑性带来的数值计算困难。最后,不同污染环境下的仿真实验结果表明算法恢复图像的信噪比、峰值信噪比、图像相似度评价值均优于比较算法恢复结果。
In order to ensure that the images including more texture information could be restored well,one image restoration method with a combined non-convex non-smooth regularization is presented.On the one hand,the first order non-convex non-smooth regularization is used to restore the images and protect the texture information.On the other hand,the second order non-convex non-smooth regularization is utilized to reduce the layering effect caused by the first order non-convex non-smooth regularization.The alternative direction method,etc are used to overcome the numerical difficulties caused by the non-convexness and non-smoothness.Finally,the simulation experiment results show that the SNR,PSNR and SSIM of the images restored by proposed method are all better than the results of comparing algorithms under different blurred conditions.
作者
刘晓光
高兴宝
LIU Xiao-guang;GAO Xing-bao(School of Computer Science and Technology,Southwest University for Nationalities,Chengdu 610041,China;College of Mathematics and Information Science,Shaanxi Normal University,Xi an 710062,China)
出处
《科学技术与工程》
北大核心
2018年第7期197-202,共6页
Science Technology and Engineering
基金
四川省教育厅重点项目(17ZA0414)
国家自然科学基金(61273311)
中央高校基本科研业务费项(2015NZYQN30)
四川省科技厅重点支撑项目(2016RZ0065)资助
关键词
非凸
非光滑
正则
仿真
图像恢复
non-convex
non-smooth
regularization
simulation
image restoration
