一种新的运动模糊图像恢复方法  

A new method for restoration of motion blurred images

作　　者：陈波[1] 

机构地区：[1]深圳大学数学与计算科学学院,深圳518060

出　　处：《计算机应用》2008年第8期2024-2026,2029,共4页journal of Computer Applications

基　　金：深圳大学实验室开放基金项目(2007303;2007308;2007311);深圳大学科研启动基金资助项目(200863)

摘　　要：通过对运动模糊产生原因的分析,提出了一种去运动模糊的新方法。首先应用Hough变换和自相关函数估计出运动模糊的方向和长度,然后应用迭代步长自适应的整体变分模型进行图像恢复。实验结果表明,这样的空间域处理方法,不但可以避免传统的频率域去模糊方法产生的震铃效应,而且该方法具有良好的抗噪性和对运动模糊参数估计误差的低敏感性。Through the analysis on reason why and when motion blurring produces, a new model for restoration of motion blurring image method was proposed. Hough transform and self-correlative function were first employed to estimate the direction and length of the motion blurring, and then total variation model which was time-step adaptive is applied to restore the image. Experimental results show that such a spatial domain based processing can not only avoid the ring-bell effect incurred by traditional frequency domain processing and it is also insensitive to noise and the estimation error of motion-blurred parameters.

关 键 词：图像恢复 运动模糊 HOUGH变换 自相关函数 整体变分模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]