摘要
针对高分辨率遥感卫星在大幅宽成像时,由于扫描速度提高而导致遥感图像产生运动模糊的问题,提出用长积分时间模糊图像和短积分时间无模糊图像联合去除运动模糊,获得高信噪比无模糊图像的解决方案。该方案利用低信噪比清晰图像中的细节信息来进行模糊核的估计,通过带增益控制的残差RL去卷积算法,极大地减小了振铃效应的影响;通过引入联合双边滤波器算法,有效地复原了图像的细节信息。最后通过实验分析证明该方案的良好效果。
The high resolution remote sensing satellite will have motion blur when taking wide imaging photos, because the scanning speed is accelerated. The paper proposes a method by changing the exposure time to obtain two images： the first image is an insufficient exposure non-blurred image and the second image is a sufficient exposure blurred image. Using the detail information of the insufficient exposure non-blurred image, a precise blurred kernel can be estimated. Ringing artifact can be greatly decreased by a gain-controlled Richardson-Lucy deconvolution algorithm. Detail information of the image can be restored by a joint bilateral filtering. At the end, some experiments have been made to support the effectiveness of the proposed algorithm.
作者
张广明
高爽
尹增山
李平付
ZHANG Guang-ming GAO Shuang YIN Zeng-shan LI Ping-fu(Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences ,Shanghai 200050, China Shanghai Engineering Center for Microsatellites, Shanghai 201210, China)
出处
《电子设计工程》
2017年第18期82-86,共5页
Electronic Design Engineering
基金
中科院国防科技创新基金资助项目(CXJJ-16Q133)
关键词
遥感图像
运动模糊
模糊核
振铃现象
remote sensing image
motion blur
blurred kernel
ringing artifacts
