主成分分析方法在空间干涉仪图像处理中的应用

Principal component analysis for interferometer image processing in SSRF

在干涉仪图像数据处理过程中,目前采用的多行平均图像处理算法会引入较大随机误差,且当CCD相机与狭缝之间存在小转角时,会引入较大系统误差。本文主要探究利用主成分分析(PCA)算法处理空间干涉仪图像的可行性与优势。利用MATLAB模拟空间干涉仪图像,并加入随机噪声和图像旋转,利用PCA方法和多行平均算法处理数据,比较两种算法的得到的结果误差大小。并设计CCD相机小转角实验和干涉图像弱信号实验,评估PCA算法在数据处理过程中纠正CCD相机小转角和重建弱信号图像中的效果。理论和实验均证明,PCA算法较目前多行平均算法,能更有效地降低噪声,分析弱信号图像及纠正CCD相机小转角,消除其带来的系统误差。

The interferometer system for transverse beam size measurement was one of the important parts of storage ring diagnostics in Shanghai Synchrotron Radiation Facility （SSRF）. The average of multiple lines and curve fitting method was adopted for raw image data processing now, but there were big random errors and system errors when the CCD is not aligned with the double slits. The article discusses the feasibility and the advantages about using principal component analysis（PCA） method to process the raw image data from the interferometer system. Using MATLAB simulation and two experiments with random noise and rotated CCD, comparing the results errors between PCA raw image data processing and normal average of.multiple lines method, it is concluded that the noise could be reduced and the small angle between CCD and double slits could be found out and corrected using PCA methods.