基于小波变换的磁铁矿近红外光谱数据去噪方法研究

Research on the denoising method of the magnetite near-infrared spectroscopy data based on the wavelet transform

近年来基于近红外光谱数据建模方法已成为矿物成分分析的主要方法之一,但由于各种因素的影响,在近红外光谱数据中存在大量的噪声,基于小波变换是近红外光谱数据去噪的有效方法。文中以大量的磁铁矿近红外光谱数据为数据源,利用多种小波基及软硬阈去噪方法对磁铁矿近红外光谱数据进行了去噪研究。以信噪比和均方误差为滤波效果的判定依据,结果表明硬阈去噪效果优于软阈去噪,且采用db4小波基的硬阈去噪效果最佳。

In recent years, the modeling method based on near - infrared spectral data had become one of the major methods of mineral component analysis, however, due to various factors, there was a lot of noise in the near infrared spectral data, the wavelet transform was an effective de - noising method of near infrared spectroscopy data. In this paper, taking a large number of magnets near - infrared spectroscopy data are as the data source, a variety of soft and hard wavelets threshold de - noising methods were used to study the de - noising effect of near infrared spectros- copy of magnetite, the snr and the rms error were used as a criterion for the filter effect. The results showed that the effect of the hard threshold de - noising was better than the soft threshold de - noising, and the effect of the hard threshold de - noising of db4 wavelet was the best.