小波分析用于表面等离子体波光谱的噪声滤除  

Noise-filtering of surface plasmon resonance spectra with wavelet analysis

作　　者：杜艳[1] 梁大开[1] 

机构地区：[1]南京航空航天大学航空科技智能材料与结构重点实验室,江苏南京210016

出　　处：《量子电子学报》2007年第3期294-299,共6页Chinese Journal of Quantum Electronics

基　　金：国家自然科学基金(60477031);国家自然科学基金重点项目(60535020);江苏省自然科学基金(BK2004128)资助项目

摘　　要：针对光纤表面等离子体波传感器,对小波阈值去噪的方法在测定表面等离子体波光谱中的应用进行了理论研究．首先将高斯白噪声加入理论SPR光谱,利用Daubechies9和Symlet7、11、14、15等五种小波分别对其进行去噪．将去噪后的共振峰强度和共振波长与理论值相比较,可知Symlet11小波用于SPR光谱的去噪性能较好．搭建光纤SPR传感器实验系统,利用Symlet11小波对在五个不同时刻测得的SPR反射光谱进行去噪,去噪后的共振波长和共振强度的标准偏差分别为0．4039 nm和0．2231％,通过对比证明小波阈值去噪方法能较好地修正由于噪声所产生的共振波长漂移,使得光纤SPR传感器满足实际运用的需要．Based on optical fiber surface plasma resonance sensor, the application of wavelet denoising with threshold for the SPR reflected spectrum is studied. At first, Gaussian white noise is added to the theoretical spectrum, and five kinds of mother wavelets, which are Daubechies9, Symlet7, Symlet11, Symlet14 and Symlet15 are selected to reduce the noise. Comparing the resonant reflectivity and wavelength of SPR spectrum after denoising with that of the theoretical value, Symlet11 wavelets obtained the best performance. Then, five groups of the experimental spectrum data were acquired from optical fiber SPR sensing system at different time, and their noise are reduced by Symlet11 wavelets. The standard deviations of experimental resonant wavelength and reflectivity after denoising are 0.4039 nm and 0.2231%. All these work proves that wavelet denoising with threshold＇ can well correct the migration of resonant wavelength caused by noise, and it will bring convenienceto the practical application of optical fiber SPR sensor.

关 键 词：图像与信息处理 小波分析 阈值去噪 表面等离子体共振光谱 

分 类 号：TP212.14[自动化与计算机技术—检测技术与自动化装置]