一种卫星平台振动光谱成像数据分块校正方法  

A sectioned method to correct spectral imaging data degraded by satellite vibrations

作　　者：南一冰[1] 唐义[1] 张丽君[1] 常月娥[1] 陈廷爱[1] 

机构地区：[1]北京理工大学光电学院,光电成像技术与系统教育部重点实验室,北京100081

出　　处：《物理学报》2014年第1期90-98,共9页Acta Physica Sinica

基　　金：国家重点基础研究发展计划(973计划)(批准号:2013CB329202;2009CB724005)资助的课题~~

摘　　要：卫星平台的振动会降低光谱成像数据的质量.以星载色散型成像光谱仪为例,介绍了其运动成像的退化机理.针对传统二维反卷积算法存在的问题以及推扫机理的特殊性,提出了一种光谱成像数据分块校正方法.该方法将分块处理、升维运算以及灰度渐变拼接相结合,将基于自然景物统计规律的图像去模糊算法运用于降质成像光谱数据的校正中.分别进行了不同目标的光谱成像退化校正仿真实验,实验结果表明,光谱成像数据质量在空间维和光谱维都有了明显提高,校正效果优于传统二维反卷积算法.The quality of spectral imaging data will be degraded by the vibrations of satellite platform.We study the methods to correct degraded spectral imaging data in this paper.First,the vibration patterns of satellites and the degradation mechanism of the motion imaging of dispersion imaging spectrometer are introduced.Considering the problems of traditional two-dimensional deconvolution algorithms,and the speciality of push-sweeping mechanism,a sectioned correction method is presented,in which are combined image dividing,dimension rising,and gradual splicing.And the motion deblurring algorithm based on natural image statistics is applied in the correction of degraded spectral imaging data.We implement degradation and correction experiments for various data.Results show that the qualities of spectral imaging data are signifcantly improved for both the spatials and spectrals.Our method is better than the traditional deconvolution algorithms.

关 键 词：振动 光谱成像 退化 校正 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]