基于指数律分布的快速盲图像解卷积在光电探测系统中的应用  

Blind Image Deconvolution Based on Power Law Distribution Applied in Optoelectronic Detection System

作　　者：曹雷[1,2,3,4] 陈洪斌[1,3] 邱琪[2] 任戈[1,3] 徐智勇[1,3] 张建林[1,3] 亓波[1,3] 

机构地区：[1]中国科学院光电技术研究所,四川成都610209 [2]电子科技大学光电信息学院,四川成都610054 [3]中国科学院光束控制重点实验室,四川成都610209 [4]中国科学院大学,北京100049

出　　处：《中国激光》2015年第3期276-282,共7页Chinese Journal of Lasers

基　　金：中国科学院知识创新工程(YYYJ-0815)

摘　　要：由于大气湍流影响,光电探测系统一般只能获得模糊降质图像。根据大多数图像频谱幅度谱具有指数律分布的特征,提出了一种新的快速盲图像解卷积算法,该算法基于图像谱服从指数律分布的观察,从降质图像中直接获得降质过程的初始估计,大大加快了算法收敛速度,提高了算法稳定性;以此为基础,采用加权增量维纳滤波的方法交替迭代进行点扩展函数与原始图像的进一步估计,保证了算法的良好解卷积性能和普适性,最终实现了快速有效的盲解卷积。实验表明,基于指数律的增量维纳滤波(BPL-IWF)盲图像解卷积后处理的光电探测系统,能够实时探测目标,并准实时输出BPL-IWF盲图像解卷积图像。Because of the impacts of atmospheric turbulence, only blurred degraded image is obtained with optical detection system. A novel fast blind image deconvolution algorithm is proposed based on the specialty that most images spectral amplitude is of power law distribution, and the original estimate of the degrading process directly from degraded image can largely accelerate the speed of algorithm convergence and improve the stablity. Based on the estimation, the fine estimate of point spread function is carried on, which iteratively combined with Wiener filter image deconvolution, it ensures that the algorithm is of accurate deconvolution and universality, and finally deconvolute the image fast and effective. Experimental results show that a photoelectric detection system based on power law incremental Wiener filter（BPL-IWF） blind deconvolution of images can detect targets real time and output BPL-IWF blind deconvolution image in quasi-real time.

关 键 词：图像处理 盲图像解卷积 光电探测 点扩展函数估计 准实时 增量维纳滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]