散斑干涉相位条纹图的频域滤波处理  

Frequency domain filtering for phase fringe patterns of digital speckle pattern interferometry

作　　者：王永红[1] 李骏睿[1] 孙建飞[1] 刘佩[1] 杨连祥[1,2] 

机构地区：[1]合肥工业大学仪器科学与光电学院,安徽合肥230009 [2]美国奥克兰大学机械工程系,密歇根罗切斯特48309

出　　处：《中国光学》2014年第3期389-395,共7页Chinese Optics

基　　金：国家自然科学基金资助项目(No.51075116;No.51375136);安徽省国际科技合作计划资助项目(No.12030603012);教育部留学回国人员科研启动基金资助项目(No.2011JYLH1150)

摘　　要：为了解决在数字散斑干涉技术测量时,散斑干涉相位条纹图像中大量噪声对相位解包裹结果和精度产生严重影响的问题,介绍了一种条纹正余弦分解和频域低通滤波结合的方法,实现了散斑干涉相位条纹图的高精度滤波。该方法的基本思路是在对相位图像进行滤波处理前,先将相位图通过正余弦函数进行映射转换成两幅图,分别经过频域滤波,然后再合成为相位图。这种分解频域滤波方法可以在滤波的同时,有效保留相位跳变信息。实验结果表明：与传统的图像降噪方法相比,该方法能够在保留图像＂尖峰＂信息的基础上,较好地滤除图像中的散斑噪声,方法简单有效,有效解决了传统滤波方法应用在相位条纹图中,相图灰度信息丢失10%~40%的问题。In order to solve the problem of serious influence of much noise in the phase fringe patterns of digital speckle pattern interferometry on the result and accuracy of phase unwrapping,an effective method of stripe sine / cosine decomposition combined with frequency domain low-pass filtering is introduced in this paper,and the high precision filter of phase fringe patterns of digital speckle pattern interferometry is realized. Basic idea of this method is that phase fringe patterns is mapped into two patterns by the sine / cosine function before the phase image filter processing,and then is synthesized into phase fringe patterns through frequency domain filter. This decomposition frequency domain filtering method can retention phase jump information effectively when filtering. Experimental results show that compared with the traditional image noise reduction method,this method can filter out the speckle noise in images better on the basis of reserving ＂rush＂image information,and it is simple and effective,which solves the problem of loss from 10% to 40% for phase fringe patterns gray level information when traditional filtering methods are used in stripe phase fringe patterns.

关 键 词：干涉测量 散斑干涉 相位图 频域滤波 噪声 

分 类 号：O436.1[机械工程—光学工程] TP394.1[理学—光学]