干涉条纹图像处理的相位解包新方法

APhase unwrapping algorism for image processing of interferogram

  • 摘要: 干涉图的处理是光干涉计量中的关键技术。采用泰曼-格林型干涉系统,建立被测物与干涉相位之间的数学模型,通过MATLAB软件,实现对被测参数的自动化测量。基于二维快速傅里叶方法的基本原理,提出一种新的相位解包算法菱形种子算法,通过识别1个种子点,然后依次向相邻4点扩展,再把这4个点作为第二批种子点,依次向各自的4点邻域扩展,以菱形轨迹遍历所有的有效信息点,以到达整幅图像相位解包的目的。采用该算法测量薄膜样片的厚度,测试结果与ZYGO测试结果比较,PV误差为0.0364,RMS最大误差为0.003,证明该算法虽然处理的是单幅干涉图,但可以得到高精度的相位分布。

     

    Abstract: Interferogram processing is one of main techniques in optical interferometry metrology. Modern interferometry for thin-film thickness has the advantages of non-contact, high accuracy and great field of view, etc. Taking advantage of Twyman-Green interferometer, the relationship between interferogram and measured object parameter can be determined by mathematical model, thus the thickness of measured object is measured automatically. A new algorism of phase unwrapping for measuring the material object is proposed based on the FFT method, an algorism of diamond phase unwrapping. This algorism identifies a seed point, seed points spread to four points nearby the fist seed point, the four points will serve as the second group of seed points, these seed points will spread to four points nearby the second group of seed points in turn and pass through all of the effective information points by a diamond path. Seed points will eventually bring about the phase unwrapping in the entire image. In this paper, the thin-film thickness is determined by the method, setting the results of this method against results from ZYGO, the PV error and RMS maximum error are 0.0364 and 0.002 respectively. Although single interferogram was processed in this study, a phase distribution with high accuracy was achieved.

     

/

返回文章
返回