时间序列三维荧光光谱数据量大和信息丰富的特点虽然有利于有机物的定性和定量分析, 但是大量冗余信息的存在增加了计算的复杂度和计算量。 在分析了时间序列三维荧光光谱的时频特征后, 分别利用聚类分析和二维小波变换从时间维和光谱维对三维荧光光谱进行了压缩。 在聚类分析中探讨了样本距离、 类间距离、 复合相关系数和R平方统计量等关键因素。 相关系数和R平方统计量的结合不仅提高了聚类分析的精确度, 而且减少了二维小波在光谱维进行数据压缩的工作量。 相关的数值实验表明压缩后的数据保留了原有时间序列三维荧光光谱的重要信息。

There has been abundant data saved in time series three-dimensional fluorescence spectroscopy, which is helpful to the qualitative and quantitative analysis of organic matter. However, redundant information also complicates the analysis and decreases the computation efficiency. Based on time-frequency of time series three-dimensional fluorescence spectroscopy, three-dimensional fluorescence spectroscopy is compressed with cluster analysis and 2-D wavelet transform. Some key factors, such as sample distance, inter-calss distance, composite correlation coefficient and R-square stastic, are discussed. The introductions of correlation coefficient and R-square statistic not only improve the precision of cluster analysis but also reduce the data for 2-D wavelet transform. Experiment results show that the important information in the original data is still kept in the compressed time series three-dimensional fluorescence spectroscopy.