光学遥感运动补偿压缩成像  

Motion Compensation Compressive Imaging in Optical Remote Sensing

作　　者：王泽龙[1] 朱炬波[1] 严奉霞[1] 刘吉英[1] 

机构地区：[1]国防科学技术大学数学与系统科学系,长沙410073

出　　处：《中国空间科学技术》2014年第5期56-63,共8页Chinese Space Science and Technology

基　　金：国家自然科学基金(61405251);高分辨率对地观测青年创新基金资助项目

摘　　要：针对单像素相机的运动导致压缩成像品质下降问题,提出了运动补偿压缩成像方法。首先建立运动条件下压缩采样的精确模型,进而分析了逐帧重构与联合重构两种图像稀疏方法。同时探讨了所提方法对于更加复杂运动的适应性以及所提方法成像品质的分布特性。该方法利用成像系统的运动信息重新规整压缩采样,避免了由降低积分时间或者减少采样数导致的成像品质下降。数学仿真与半物理仿真表明,运动补偿压缩成像比未经补偿成像结果具有更好的成像品质。A compressive imaging method based on motion compensation was proposed to overcome the low-quality problem resulted from the motion of single-pixel imager.Firstly,the compressive sampling model with motion condition was built.Secondly,two image reconstructions, namely, frame-by-frame recovery and joint recovery, were analyzed.Meanwhile,its adaptability to different motion types as well as the distributions of its imaging quality was also discussed.The proposed method makes use of motion information to program the samples instead of reducing integrate time or sampling number,which avoids the decrease of imaging quality. Mathematical simulations and semi-physical experiments show that the proposed method has higher imaging quality than the imaging method without motion compensation.

关 键 词：压缩成像 单像素 运动补偿 稀疏约束重构 光学相机 航天遥感 

分 类 号：TP722[自动化与计算机技术—检测技术与自动化装置]