中国激光, 2014, 41 (9): 0913001, 网络出版: 2014-08-12
基于多帧湍流退化图像的近视解卷积复原
Myopic Deconvolution Restoration Based on Multiframe Turbulence Degraded Images
图像处理 大气湍流效应 光学成像 近视解卷积 图像匹配 image processing atmosphere turbulence effect optical imaging myopic deconvolution image registration
摘要
根据贝叶斯和Parseval定理,引入了频域内多帧湍流退化图像的近视解卷积复原算法。以大气湍流长曝光光学传递函数作为估计的光学传递函数。根据频域代价函数的特点,提出分步牛顿法求解代价函数。本算法能够处理未匹配的多帧图像,并能获得理想的复原图像。计算机仿真多帧湍流退化图像的复原结果表明:即使多帧图像未匹配,在不同湍流强度和不同噪声情况下算法仍能复原出好的图像效果。
Abstract
According to the Bayes and Parseval theorem, myopic deconvolution algorithm for multiframe turbulence degraded images in frequency domain is presented. Atmosphere turbulence long exposure optical transfer function is used as the estimated optical transfer function. According to characteristics of the cost function in frequency domain, partial Newton algorithm is introduced. This algorithm can handle unregistration of multiframe images and obtain ideal recovered image. The restored result of the computer simulation multiframe turbulence degraded images shows that it has good restoration effect at different turbulence intensities and signal noises, even though the images are unregistration.
冯晓星, 时东锋, 乔春红, 张鹏飞, 范承玉, 王英俭. 基于多帧湍流退化图像的近视解卷积复原[J]. 中国激光, 2014, 41(9): 0913001. Feng Xiaoxing, Shi Dongfeng, Qiao Chunhong, Zhang Pengfei, Fan Chengyu, Wang Yingjian. Myopic Deconvolution Restoration Based on Multiframe Turbulence Degraded Images[J]. Chinese Journal of Lasers, 2014, 41(9): 0913001.