基于Poisson模型的湍流退化图像多帧迭代复原算法

随着航天空间技术的发展，空中目标的成像探测研究越来越重要。受大气湍流的干扰，观测到的目标图像是严重模糊的。为了从观察到的多帧含噪的湍流退化图像中将目标原图像有效地恢复出来，本文提出了一种新颖的基于图像统计模型的图像复原算法。跟据图像Poisson概率模型建立了有关多帧图像数据的对数似然函数，通过极大化该对数似然函数，推导出了目标图像及各帧图像点扩展函数的迭代求解关系。同时。将点扩展函数的支持域等先验条件有效地加入到迭代计算过程中，以便快速地利用迭代技术将目标图像和各帧点扩展函数估计出来。该算法能用少数帧图像极大程度地恢复出目标图像。为了验证本文算法的恢复效果和可靠性，对在强噪声污染条件下的湍流退化图像进行了恢复实验，实验结果表明本文算法对空中目标湍流退化图像的复原是非常有效的。

Iterative multi-frame restoration algorithm of turbulence-degraded images based on poisson model

Because of the images matching position in INS/SAR integrated navigation system needing the unequal matching calculation time,which resulted in in-coordinate interval and delay characters of measurement output.The integrated filtering with the common kalman filtering algorithms couldn't get high degree of accuracy for the problem.At first,the paper analyzed the common kalman filtering work process.And then the paper gave the filtering algorithms for solving the problem of in-coordinate interval and measurement delay.The paper utilized the character of the system state transition matrix,and designed the corresponding in-coordinate interval kalman filtering algorithms to solve the problem of in-coordinate interval measurement.And based on the in-coordinate interval kalman filtering algorithms,the paper gave the scheme of solving the measurement delay.With the analysis of the covariance,the paper analyzed the filtering accuracy of all of the common kalman filtering,the in-coordinate interval kalman filtering and the algorithms of solving the measurement delay.And the simulation results have verified the high degree of accuracy of the algorithms presented in paper.