基于数值保真项优化的TDI遥感图像复原方法  

Restoration method of TDI remote sensing image based on optimization of numerical fidelity term

作　　者：苏慧 冯华君[1] 徐之海[1] 李奇[1] 陈跃庭[1] SU Hui;FENG Hua-jun;XU Zhi-hai;LI Qi;CHEN Yue-ting(State Key Laboratory of Modern Optical Instrumentation,Zhejiang University,Hangzhou 310027,China)

机构地区：[1]浙江大学现代光学仪器国家重点实验室,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2018年第4期674-679,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(61475135)

摘　　要：提出基于数值保真项优化的时间延迟积分电荷耦合元件(TDI-CCD)振动模糊遥感图像复原方法.针对时间延迟积分(TDI)退化图像点扩散函数(PSF)的空间变化性,根据成像平台颤振信息逐行构建模糊核.通过结合L1正则化和数值保真项优化,提出新的遥感图像去模糊方法,实现对TDI颤振图像的逐行复原.在基于最大后验概率(MAP)构建图像去模糊模型的过程中,将数值保真项扩展到二阶,以更好地保留图像细节.与传统的RichardsonLucy(RL)算法、全变分方法相比,提出的算法去模糊效果明显.分析在不同积分级数和颤振频率下采用该算法复原后的图像的ssim,该算法体现出很好的普适性.A restoration method was proposed based on the optimization of numerical fidelity term in order to recover the time delay integration(TDI)charge-coupled device(CCD)remote sensing image with vibration blur.The blur kernel was constructed line by line according to the flutter information of the imaging platform due to the spatial variation of the point spread function(PSF)of the TDI degraded images.A new deblurring algorithm for remote sensing images was presented to restore blurred images line by line by combining L1 norm regularization and optimization in numerical fidelity.The numerical fidelity term was extended to the second order in order to better recover image details in the process of constructing the image deblurring model based on maximum a posteriori probability(MAP).The proposed algorithm obtained sharper deblurring results compared with traditional Richardson-Lucy(RL)algorithm and the total variational method.Image structural similarity index measurement(SSIM)in different integral series and vibration frequencies showed the universality of the new algorithm.

关 键 词：遥感 平台颤振 图像复原 数值保真项 时间延迟积分电荷耦合元件(TDICCD) 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]