摘要
相较于低轨卫星线阵扫描成像模式,地球静止轨道面阵成像的曝光时间相对更长,更容易受到平台颤振的影响而造成图像模糊.为了消除由平台颤振引起的像质退化,本文提出了基于分时积分亚像元融合的方法.由于地球静止轨道的凝视成像特性,相机观察区域在长时间内保持不变,因此分时短曝光可以获得多帧目标内容相同,但模糊尺度更低的短曝光图像.然后对多帧短曝光图像采用基于能量区域质心法的相位相关算法进行亚像元图像配准,计算相对偏移量并进行补偿,位移探测准确度可达0.1像元以内,满足卫星平台应用需求.再按亚像元偏移量对多帧图像进行融合,融合的过程可以提升由于曝光时间缩短而降低的单帧图像信噪比,最终可以获得图像清晰度更高、信噪比与原长曝光图像相当、信息辨识度更好的遥感图像.
Comparing to the linear scanning space camera in low satellite orbit,images captured by the area array staring imaging system in geostationary orbit usually emerges more blur because of the satellite vibration in the relative longer exposure time.To eliminate the motion blur degradation,the paper proposed to capture multiple short-exposure images and fused them on sub-pixel scale instead.The imaging system in geostationary orbit includes the feature of staring at the same area for a long time.Therefore the task mode of time-sharing short-exposure could get multiple less blurred images of the same area.Next,the sub-pixel displacements between the different frames are estimated by image registration algorithm with 0.1 pixel accuracy.And then,the images are fused to improve the low signal-to-noise ratio because of the short-exposure.Excellent results have been obtained by the sub-pixel image registration and image fusion in the paper.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2012年第11期1359-1364,共6页
Acta Photonica Sinica
基金
国家自然科学基金(No.61036015)重点项目资助
关键词
遥感图像处理
地球静止轨道
振动探测
图像配准
图像融合
Remote sensing
Geostationary orbit
Vibration estimation
Image registration
Image fusion
