基于频域的运动模糊参数估计  

Estimation of Parameters for Motion-blurred Image Based on Frequency Domain

作　　者：张永梅[1] 张静[1] 

机构地区：[1]北方工业大学信息工程学院,北京100144

出　　处：《电脑开发与应用》2014年第7期1-5,共5页Computer Development & Applications

基　　金：国家自然科学基金资助项目(61371143);国家科技支撑计划项目(2012BAH04F00);北京市自然基金资助项目(4132026)

摘　　要：运动模糊图像复原的关键之一是模糊核函数的估计,即降质图像的模糊方向和距离变化的估计。已经有多种方法用于模糊核函数的估计,通过对运动模糊图像频谱的研究,提出了一种基于形态学的模糊运动方向计算方法,并根据模糊图像频谱的黑色条纹周期出现的规律,采用投影法计算极值点,从而获取模糊运动距离。实验结果表明该算法简单有效,得到的运动模糊参数较准确。The estimation of fuzzy kernel function is one of the key factors in motion blurred image restoration,namely the estimation of blur orientation and length of the degraded image.A variety of methods has been applied to the estimation of fuzzy kernel function.By studying the spectrum of motion blurred images,this paper proposes a method to compute blur orientation based on morphology and calculates blur movement length using the projection method to get the extreme points,according to the regular pattern that black stripes periodically appear in the spectrum of blurred images.Experimental results show the algorithm is simple and effective.It is reliable for the motion blur parameters.

关 键 词：运动模糊 模糊方向 运动距离 频谱 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]