摘要
为了从湍流序列退化图像中有效地去除模糊、抖动等气动光学效应,提出一种基于内外循环迭代的湍流退化图像复原新算法.该算法利用多帧湍流退化图像,建立了一个基于航天图像随机场概率模型的联合对数似然函数.通过极大化该对数似然函数,推导出了基于内外循环交替求解目标图像及各帧点扩展函数的迭代关系,据此可将目标图像和各帧点扩展函数同时估计出来.在微机上进行了复原实验,实验结果表明该算法能用多帧图像获取目标图像和点扩展函数的最佳联合估计,可有效地去除模糊、抖动等气动光学效应.
A novel restoration algorithm based on inside and outside circulation iterations was proposed to remove efficiently the aero-optical effects including blur, dither, and so on, from a sequence of turbulencedegraded images. A united logarithmic likelihood function based on the statistic probability model of image stochastic filed was built by using the multi-frame images. The iterative relations to calculate the object image and turbulence＇s point spread functions alternately were derived by maximizing the likelihood function, estimating the object image and the point spread functions. The restoration experiments have been made on PC, showing that the proposed algorithm can obtain the optimum estimations of the object image and the point spread functions using multi-frame images, with the aero-optical effects such as blur and dither being removed efficiently
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2005年第9期15-18,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金重点资助项目(60135020).
关键词
气动光学效应
湍流退化图像
图像复原
最大似然估计
aero-optical effects
turbulence-degraded image
image restoration
maximum-likelihood estimation
