一种改进的基于Radon变换的运动模糊图像参数估计方法  

Improved approach to motion blur identification based on Radon transform

作　　者：乐翔[1] 程建[1] 李民[1,2] 

机构地区：[1]电子科技大学电子工程学院,四川成都611731 [2]桂林空军学院科研部,广西桂林541003

出　　处：《红外与激光工程》2011年第5期963-969,共7页Infrared and Laser Engineering

基　　金：国家973项目(2007CB714406);遥感科学国家重点实验室开放基金项目(2009KFJJ004);中国博士后基金(20080441198)

摘　　要：运动模糊是导致图像降质的最常见因素之一。此类图像的复原很大程度上取决于对点扩散函数的准确估计。对于匀速直线运动的模糊图像,运动方向和模糊尺度是确定点扩散函数的两个关键参数。介绍了Radon变换的数学原理及用其估计运动模糊方向的思路、步骤;针对实际拍摄的模糊图像频谱中严重影响原算法的十字亮线现象,分析了其成因,并提出了一种改进算法。算法根据十字亮线的特性,采用频谱分块的思想以消除检测干扰。实验结果验证了算法的正确性和有效性。Motion blur is one of the most common blurs that degrades images.The restoration of motion-blurred images is highly dependent on accurate estimation of the point spread function.For uniform linear motion blurred images,the motion direction and blur extent are two key parameters determining the point spread function.The mathematic principle of Radon transform and procedures of using it to identify the motion direction were introduced.Aiming at the dominant cross-shaped bright lines showing up in the Fourier spectra of the real blurred images taken by digital camera,the cause of formation was analyzed and an improved algorithm was proposed.According to the properties of the bright lines,sub-blocks method was adopted to avoid detection interference.Experimental results show the validity and effectiveness of the presented method.

关 键 词：图像处理 先验模糊辨识 RADON变换 点扩散函数 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]