模糊遥感图像总变分优化恢复方法  

A Restoration Method of Blurred Remote Sensing Images Based on Total Variation Optimization

作　　者：秦世引[1] 魏晓明[1] 刘远民[1] 满益云[2] 

机构地区：[1]北京航空航天大学自动化科学与电气工程学院,北京100191 [2]北京空间机电研究所.北京100076

出　　处：《科技导报》2011年第14期26-32,共7页Science & Technology Review

基　　金：中国高技术研究发展计划(863计划)重点项目(2008AA12A216);国家自然科学基金项目(60875072)

摘　　要：针对遥感图像在成像与传输过程中的退化而导致的图像模糊与噪声干扰,提出了一种基于调制传递函数(MTF)估算与总变分优化的图像恢复方法。通过分析调制传递函数曲线的幅频特性,阐释了基于MTF的遥感图像恢复机制,为抑制噪声干扰和扩大适用范围而采用了一种新的MTF曲线分段逼近策略,并以实验结果验证了该方法较之常规模型的优越性。进而,针对噪声条件下模糊遥感图像的恢复问题,采用改进的快速总变分优化方法实现了对强噪声模糊遥感图像的有效恢复。实验结果表明,本文提出的恢复方法能够在有效抑制噪声的同时更好地保留图像细节和纹理信息,为强噪声干扰模糊遥感图像的恢复处理提供了高效的解决方法。In view of blurred remote sensing images with the noise resulted in some degradations during imaging and transmission processes, a kind of restoration method is proposed based on MTF estimation and total variation optimization. The restoration principle of blurred remote sensing images with noise is elucidated in depth through a thorough analysis of frequency characteristics of MTF. In order to deal with the noise, a new piecewise approximation strategy for MTF estimation is employed, which has the general applicability and outperforms the conventional models in overcoming noise disturbance by experiment demonstrations. Moreover, aiming at the restoration processing of blurred remote sensing images with noise, an improved method of fast total variation optimization is adopted to achieve the optimal restoration effect. A series of experimental results indicate that proposed restoration method could preserve more details and texture information of images besides its satisfactory performance of eliminating noise, thus it could provide an effective approach to the restoration processing of remote sensing images with heavy noise.

关 键 词：图像复原 调制传递函数 总变分优化 去噪 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]