将数学中的二元凸函数判定和数据挖掘中的聚类分析方法结合,提出了针对三维荧光的光谱区域选择方法,并利用此种方法从光谱图中提取出含有丰富光谱信息的凸集区域.对水体中总有机碳的检测和白酒中黄曲霉素的检测进行了实验研究,实验结果表明,采用本文提出的三维荧光光谱区域选择方法提高了模型的精度,与利用全光谱所建立的回归模型相比,模型精度分别提高了6.17%和4.97%.“,”Based on combination of binary convex function discriminant theorem with clustering analysis,a new method of characteristic region selection for three-dimensional fluorescence spectrometry is proposed.By this method,the convex regions with valid spectral data are obtained.Experiment for detecting total organic carbon(TOC) in water and AFB1 in liquor are carried out.The experimental results show that the proposed method improves the accuracy of the regress model with increase of 6.17%and 4.97%respectively.