大尺度三维视觉测量中的离焦模糊图像恢复  

Defocused blur image restoration in large scale 3D vision measurement

作　　者：张益昕[1] 王顺[1] 张旭苹[1] 

机构地区：[1]南京大学光通信工程研究中心,南京210008

出　　处：《仪器仪表学报》2010年第12期2748-2753,共6页Chinese Journal of Scientific Instrument

基　　金：江苏省自然科学基金(BK2009257)资助项目;上海铁路科研开发与示范性推广项目(2009057)的资助

摘　　要：为了实现大型设备的几何量测量,需要对多幅立体图像进行拼接。由于镜头的焦距固定,当观测不同距离的场景时往往使图像产生离焦模糊,从而丧失图像细节,降低了拼接精度。提出了一种基于维纳滤波的离焦模糊图像恢复算法。利用模糊图像频谱上的暗环半径估计退化函数的参数,并通过对原始图像的边界进行处理消除了图像边界条件的不连续性。实验结果显示该算法可以有效地降低图像恢复过程中产生的振铃现象,对离焦模糊图像的恢复质量良好。In order to realize geometric quantity measurement of large scale equipment,a stereo image mosaic method is needed.Unfortunately,the lens usually has fixed focal length,which causes defocused blue image while observing scenes in different distances,which makes the captured image lose its high frequency feature and limits the resolution of image mosaic.In this paper,a Wiener filter based algorithm for defocused blur image restoration is presented.Blur identification with dark loop searching in frequency dimension is proposed.The boundary of image is symmetrized to reduce the ringing effects during the process of image restoration.Experiment results indicate that the new method effectively reduces the ringing effects and restores the defocused blur image effectively.

关 键 词：模糊识别 离焦模糊 维纳滤波 振铃现象 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]