运动模糊图像PSF参数估计与图像复原研究  

Research on PSF parameter estimation and image restoration of motion blurred image

作　　者：廖秋香[1] 卢在盛[1] 彭金虎[1] Liao Qiuxiang;Lu Zaisheng;Peng Jinhu(Guangxi Colleges and Universities Key Laboratory of Image Processing and IntelligentInformation Systems, Wuzhou University, Wuzhou 543002)

机构地区：[1]梧州学院广西高校图像处理与智能信息系统实验室,梧州543002

出　　处：《高技术通讯》2019年第4期338-343,共6页Chinese High Technology Letters

基　　金：国家自然科学基金(61562074);2018年广西高校中青年教师基础能力提升(2018KY0542);梧州学院重点科研(2017B006);梧州学院中青年骨干教师培养工程资助项目

摘　　要：运动模糊图像复原对于改善图像质量有重要的理论意义和现实意义。在研究运动模糊图像复原中,对点扩散函数(PSF)的估计是关键点也是难点。本文利用Radon变换原理来求解点扩散函数PSF中的运动模糊方向,并提出了消除十字亮线引起的干扰的新方法。利用运动模糊图像频谱上的中心暗条纹间距来计算运动模糊尺度。基于估计的PSF参数采用维纳滤波算法来恢复运动模糊图像。实验结果表明,运动模糊参数估计精确,运动模糊方向控制在1°以下,运动模糊尺度控制在1个像素以内。同时采用维纳滤波算法来恢复运动模糊图像,效果优异,可获得细节清晰的图像。The restoration of motion blurred images has important theoretical and practical significance for improving the quality of the image. Estimation of point spread function(PSF) is crucial and difficult in research on restoration of motion blurred images. The motion blur direction in point spread function is gained by using the Radon transform principle, and a new method to eliminate the interference caused by the cross line is proposed. The motion blur extent is calculated by using the central dark fringe distance on the motion blurred image spectrum. The Wiener filtering algorithm is used to restore the motion blurred image based on the estimated PSF parameters. The experiment results show that the motion blur parameters are estimated accurately, the estimated error in blurred direction is less than 1 degree, and the estimated error of blurred extent is less than 1 pixel. At the same time, the restoration based on the Wiener filtering algorithm has good restore effect and gain clear details.

关 键 词：点扩散函数(PSF) 模糊方向 模糊尺度 RADON变换 维纳滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]