多菌灵农药激光拉曼光谱的小波去噪方法研究  

Study for wavelet denoising methods of laser Raman spectrum of carbendazim pesticide

作　　者：王晓彬[1] 吴瑞梅[1,2] 刘木华[1] 蔺磊[1] 张庐陵[1] 

机构地区：[1]江西农业大学工学院,江西南昌330045 [2]江苏大学农产品物理加工重点实验室,江苏镇江212013

出　　处：《食品工业科技》2014年第5期277-280,共4页Science and Technology of Food Industry

基　　金：国家自然科学基金项目(31271612);江西省自然基金项目(20122BAB204020);江西省科技攻关项目(2011BDH80010);江西省教育厅科技项目(GJJ13272);江苏省农产品物理加工重点实验室开放项目(JAPP2012-3)资助

摘　　要：利用激光拉曼光谱仪采集多菌灵农药标准品拉曼信号,分别采用移动平均平滑法、小波软阈值法和小波硬阈值法对光谱信号进行去噪处理,对比各方法的去噪效果。结果表明:小波硬阈值法可获得最优去噪质量,当小波基函数为sym2,尺度分解为5,阈值量化为‘Heursure’,硬阈值处理,重构光谱的信噪比最大(为60.927),均方根误差最小(为11.429)。研究表明:小波硬阈值法能有效去除多菌灵农药拉曼光谱信号的噪声,最大程度保留拉曼光谱特征信息,为食品及农产品中农药残留的拉曼光谱快速检测提供方法支持。Raman signal of carbendazim pesticide was collected by laser Raman spectrometer. Moving-average smoothing method, wavelet soft-threshold method and wavelet hard-threshold were applied respectively to denoise the acquired Raman signals,and their denoising effects were compared.The results showed that wavelet hard-threshold could obtain the optimal denoising quality. When wavelet base function being sym2, scale decomposition being 5, the threshold quantization being ＇ Heursure＇, hard- threshold value being processed, signal to noise ratio（SNR）of the reconstructed spectral was maximum which was 60.92.7, and root mean square error（RMSE） was minimum which was 11.429.The study showed that wavelet hard-threshold could effectively remove the noise information of the raw Raman signals of carbendazim pesticide,and saved the most number of spectra details.The study could provide a methodological support for the rapid detection of pesticide residue in food and agricultural products based on Raman spectrum.

关 键 词：多菌灵农药 激光拉曼光谱 小波分析 去噪 

分 类 号：TS207.7[轻工技术与工程—食品科学]