小波变换分光光度法可行性研究(Ⅰ.理论研究)  

Research into Computational Spectrophotometry Based on Wavelet Transform (part Ⅰ.Theoretically Study)

作　　者：王艳[1,2] 相秉仁[1] 

机构地区：[1]中国药科大学分析计算中心 [2]中南大学化学化工学院 长沙 410083

出　　处：《药物分析杂志》2005年第9期1149-1153,共5页Chinese Journal of Pharmaceutical Analysis

摘　　要：目的:从理论上研究基于小波变换的计算分光光度法的可行性。方法:用高斯函数模拟物质的吸收光谱,用34个小波函数进行小波变换,提取只与待测组分含量有关的特征小波系数。结果:连续小波变换比离散小波变换性能好,采用连续小波变换处理两组分重叠吸收光谱数据,不同的小波函数性能差别很大,不存在对所有情况都适用的最佳小波,但只要干扰物与待测物的光谱差异满足一定的条件,即两者的最大吸收波长不重合,或者两者的最大吸收波长重合,但峰宽相差不小于2nm,通过选择合适的小波函数可以得到只与待测物浓度相关的特征小波系数。结论:理论研究表明基于小波变换的计算分光光度法可行。Objective：To study the feasibility of computational spectrophotometry based on wavelet transform theoretically. Method：Gaussian peaks with different parameters of peak position, width and height are used to simulate the absorption spectra. Absorption spectra were manipulated by wavelet transform to obtain eigen wavelet coefficients that only correlated with the concentration of the component to be measured. 34 wavelets were investigated. Results： Continuous wavelet transform is better than discrete wavelet transform. If the overlapping absorption spectrum data of two components are processed by continuous wavelet transform, eigen wavelet coefficient could be obtained with the relative error less than 2% , so far as the peak position of the two overlapping absorption spectra is not the same, or they are the same, but the difference between peak width is not less than 2nm. In this process,34 wavelets behavior quite different from each other, and there doesnt exist an optimum wavelet function that can be applied to transform all overlapping absorption spectrum data of two components to obtain eigen wavelet coefficient. Conclusion：continuous wavelet transform can be utilized theoretically as a novel method of data processing in computational spectrophotometry.

关 键 词：小波变换 连续小波变换 计算分光光度法 最大吸收波长 吸收光谱 小波函数 组分含量 高斯函数 光谱数据 干扰物 

分 类 号：R241.1[医药卫生—中医诊断学] R971.2[医药卫生—中医临床基础]