您的位置:首页  > 论文页面

基于边缘和纹理信息的运动模糊图像盲复原算法

发表时间:2015-03-31  浏览量:1585  下载量:808
全部作者: 孙宇恒,胡绍海,刘帅奇
作者单位: 北京交通大学信息科学研究所;河北大学电子信息工程学院
摘 要: 提出一种基于边缘和纹理信息的运动模糊图像盲复原算法,该方法充分利用自然图像中存在的各种先验知识对模糊图像进行复原,分为模糊核估计和图像修复2个阶段。在模糊核估计时,采用双边滤波器和冲击滤波器从模糊图像中预测出清晰边缘,以此指导全局图像的复原,并运用多尺度策略解决大模糊核问题。在图像修复时,采用基于边缘和纹理信息的分裂Bregman迭代算法复原出清晰的图像。实验结果表明,算法有效解决了运动模糊图像盲复原问题,复原结果既保留了更多的细节信息,又抑制了振铃效应,而且对噪声具备较好的鲁棒性。
关 键 词: 图像处理;运动模糊;盲复原;边缘信息;纹理信息;分裂Bregman迭代算法
Title: A blind motion deblurring algorithm based on edge and texture information
Author: SUN Yuheng, HU Shaohai, LIU Shuaiqi
Organization: Institute of Information Science, Beijing Jiaotong University; College of Electronic and Information Engineering,Hebei University
Abstract: A blind motion deblurring algorithm based on edge and texture information is proposed, which makes full use of nature image priors. It contains two phases: kernel estimation and image restoration. In the procedure of kernel estimation, bilateral filter and shock filter are used for predicting clear edge from the blurred image, which conducts global image restoration. Furthermore, multi-scale scheme is applied to solve big blurred kernel problem. In the step of image restoration, split Bregman iteration method based on edge and texture information is utilized to restore clear image. Experimental results show that the proposed algorithm can effectively solve blind motion blurring problem. The restoration result can not only reserve more details, but also suppress ringing effects effectively and is robust to noise.
Key words: image processing; motion blurring; blind deblurring; edge information; texture information; split Bregman iteration method
发表期数: 2015年3月第6期
引用格式: 孙宇恒,胡绍海,刘帅奇. 基于边缘和纹理信息的运动模糊图像盲复原算法[J]. 中国科技论文在线精品论文,2015,8(6):626-634.
 
0 评论数 0
暂无评论
友情链接