HHT时频分析土壤光谱的重金属铜离子污染信息提取模型

摘要

The amount of information that the different Cu2+contents in soil are mapped to the soil spectra is very weak,and the noises in the hyperspectral data are also very difficult to avoid.Thus the sky point of the research would be how to extract the weak Cu2+information from the complex noise environment of soil spectra.The empirical mode decomposition(EMD)algorithm can effectively remove the noise in hyperspectral data.Further more,the EMD is the premise of Hilbert transform that is a kind of time-frequency analysis on nonlinear and unstable signal.Therefore the time-frequency analysis algorithm on Hilbert-Huang transform(HHT)could be used for de-noise processing and information extracting of the soil spectra after the Huang transform is introduced.In this paper,through the HHT was applied in time-frequency to analyze the soil spectra polluted by different Cu2+concentrations,and the information mining of Cu2+pollution in soil spectra was achieved based on the curves such as enve-lope,modulation signal and frequency spectrum of each intrinsic mode function(IM F)component that could be decomposed out from the original soil spectra by EMD.The study results showed that,the analyzing results of soil spectra were identical based on the time-frequency HHT when the soil was polluted by same Cu2+concentrations,but non-identical when polluted by differ-ent Cu2+concentrations,so that the Cu2+content in soil can be also retrieved based on the IM F components.Accordingly,the HHT analysis in time-frequency provides a new method and idea for information extracting,spectral diagnosing and Cu2+content retrieving of soil spectra.%土壤中不同浓度Cu2+含量映射到土壤光谱上的信息量十分微弱,并且这些高光谱数据中也存在着难以避免的噪声,因而本研究的关键是如何在土壤光谱复杂的噪声环境中提取微弱Cu2+信息.经验模态分解算法(EMD)能够有效去除高光谱数据中的噪声,且EMD是Hilbert变换对"非线性非稳定"信号时频分析的前提,当引入Huang变换后,可利用Hilbert-Huang变换(HHT)模型时频分析高光谱数据以实现降噪处理与信息提取.通过时频的HHT分析不同浓度Cu2+污染下的土壤光谱,完成从原始光谱经EMD分解出各本征模态函数(IM F)分量的包络线、调制信号和频谱等曲线中挖掘土壤光谱的Cu2+污染信息.研究结果表明,相同浓度Cu2+污染时的土壤光谱HHT时频分析结果相同,不同浓度时则不同,所以也可依据IMF分量反演土壤Cu2+含量.因此,高光谱数据的HHT时频分析能为土壤光谱的信息挖掘、光谱诊断和Cu2+含量反演等提供一种新的方法和思路.