基于MATLAB GUI的谱线平滑处理

Research on the spectrum smoothing method based on MATLAB GUI

谱线平滑可以去除实测核能谱中的统计涨落并保留谱峰重要信息。首先介绍了高斯平滑和SavitzkyGolay(SG)平滑的原理与实现方法,然后基于矩阵实验室图形界面开发环境(Matrix Laboratory Graphical User Interface,MATLAB GUI)设计了谱线分析工具,对谱数据进行在线实时和离线批量处理。最后对石墨预衍射X荧光分析仪实测的U标准样数据作平滑分析,并引入相关系数来评价平滑效果。实验结果表明:当谱峰幅值较大时,高斯平滑的平滑度指标较好,当谱峰幅值较小时,SG平滑显现优势;高斯平滑的参数设置较灵活,但运算较复杂,需用到矩阵计算,SG平滑的算法较简洁,代码执行时间更短。因此,实际工程应用中,应综合考虑在维持较好平滑度和相关系数指标的情况下,合理选择平滑方法和参数。

[Background] Spectrum smoothing can remove the statistical fluctuations in the measured energy spectrum and retain important information of the spectral peaks. Each spectrum smoothing method has its own advantages, and is not always effective in any case. [Purpose] This study aims at the application of spectrum smoothing methods with different smoothing times and parameters, and evaluate smoothing methods based on smoothness and correlation coefficient respectively. [Methods] A spectrum analysis tool which is integrated Gauss smoothing and Savitzky-Golay（SG） methods is designed based on Matrix Laboratory Graphical User Interface,（MATLAB GUI）. It is applicable to real-time and off-line batch processing of spectral data with a few necessary steps. One spectrum captured by the graphite pre-diffraction X fluorescence analyzer is processed by the designed tool, and the correlation coefficient is introduced to evaluate the smoothness. [Results] Experimental results show that the smoothness of the Gauss smoothing is better than SG when the peak amplitude is larger. Whilst SG outperform Gauss smoothing when the peak amplitude is small. The Gauss smoothing parameters are more flexible than that of SG, but the operation is more complex that needs matrix manipulation. The SG smoothing algorithm is simple and the execution time of the code is short. [Conclusion] The smoothing methods and parameters should be selected in accordance with actual conditions to maintain good smoothness and correlation coefficient for spectrum smoothing.