运动模糊车牌图像的复原算法研究与实现  

Study and Realization for Image Restoration of Motion Blur Vehicle

作　　者：黄小芬[1] HUANG Xiao-fen(Concord College, Fujian Normal University, Fuzhou Fujian, 35011)

机构地区：[1]福建师范大学协和学院,福建福州350117

出　　处：《山西大同大学学报（自然科学版）》2018年第3期35-38,共4页Journal of Shanxi Datong University(Natural Science Edition)

基　　金：福建省教育厅科技项目[JAT160668]

摘　　要：采用模型估计法,由车辆匀速直线运动的先验知识推导出模糊退化传递函数。在运动模糊参数估计中,通过二次傅里叶变换得到模糊图像频谱,采用Radon变换检查直线方法检测出模糊角度,并用旋转投影法估算出运动尺度,最后用约束的最小二乘方滤波复原法复原出清晰的车牌图像。通过Matlab软件仿真,对该算法与其他复原算法从执行效果和复原质量进行对比实验。结果表明:该算法不仅能准确的估计出模糊角度和尺度,也能快速的复原出清晰的车牌图像。The fuzzy degenerate transfer function is deduced from the prior knowledge of the uniform motion of the vehicle by the model estimation method first. Then, in motion fuzzy parameter estimation, the fuzzy image spectrum is obtained by the second Fouriertransform, the blur angle is detected from the straight line method of radon transforms, and the rotation projection method is used to esti？mate the movement scale. Finally, a clear license plate image is restored by using the constrained least squares filter recovery method.Using programming method of Matlab, the algorithm is compared with other restoration algorithms from the implementation effect andthe restoration quality. The results show that, the algorithm can not only accurately estimate the motion blurred direction and motionblurred length, but also quickly restore a clear license plate image.

关 键 词：图像复原 参数估计 二次傅里叶变换 约束最小二乘法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]