Radon变换对短模糊尺度下匀速直线运动模糊参数的准确估计  

Accurate Estimation of Motion Blur Parameters at Short Blurred Extent Based on Radon Transform

作　　者：段若颖[1] 谌德荣[1] 蒋玉萍 关咏梅 高翔霄 

机构地区：[1]北京理工大学机电工程与控制国家重点实验室,北京100081 [2]北京宇航系统工程研究所,北京100076

出　　处：《兵工学报》2013年第10期1231-1235,共5页Acta Armamentarii

基　　金："十二五"国防基础科研基金项目(B2220132013)

摘　　要：为提高短模糊尺度下的运动模糊参数估计精度，提出了一种改进的基于Radon变换的频域估计算法。算法对运动模糊图像傅里叶变换频谱图做Radon变换，利用运动模糊图像频谱图在垂直于条纹方向下Radon变换时为明暗投影线的特性，在Radon变换图中确定明暗投影线，由该投影线对应的Radon变换角度和运动模糊图像的分辨率求解运动模糊参数。对Lena图像的仿真结果表明：当模糊尺度为5～10时，对模糊方向的估计误差小于1°，对模糊尺度的估计误差小于0．5个像素，优于现有算法的估计精度。An improved algorithm based on Radon transform is proposed to identify the motion blur pa- rameters of image at short blurred extent more accurately. Considering that the spectrogram of motion blurred image is transformed into an alternate dark and bright projection line at perpendicular direction of dark stripe by Radon transform, the position of alternate dark and bright projection line in Radon trans- form image should be determined. The motion blur parameters can be determined by the Radon transform angle which corresponds to the projection line and the resolution of motion blurred image. The experimen- tal results show that the estimated error in blurred direction is less than 1 ° and the estimated error of blurred extent is less than 0.5 pixels when the blurred extent is 5 - 10 pixels. The estimation accuracy of improved algorithm is higher than other algorithm at short blurred extent.

关 键 词：兵器科学与技术 短模糊尺度 RADON变换 运动模糊参数 绝对误差 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]