提升小波和Savitzky-Golay滤波结合的近红外光谱去噪研究

Near Infrared Spectrum Denoising Based on Lifting Wavelet Transform and Savitzky-Golay Filter

在近红外光谱数据处理中,测得的近红外光谱数据不仅有被测样品的近红外特征光谱,还包含一些随机噪声,噪声的存在会影响后续光谱分析的准确性,为提高近红外光谱分析精度,需要对近红外光谱数据进行去噪处理。单一的提升小波去噪、小波去噪以及Savitzky-Golay滤波很难获得较好的去噪效果。因此提出将提升小波变换结合Savitzky-Golay滤波方法用于近红外光谱去噪,对降噪效果进行仿真与评估并与单一去噪方法进行对比。分别对添加随机噪声的1467 nm近红外光谱进行单一小波去噪、提升小波去噪、Savitzky-Golay滤波以及提升小波变换结合Savitzky-Golay滤波进行去噪。实验结果显示所提出的方法去噪后的信噪比比单一三种去噪方法分别提高0.3364、1.1074、0.1287,均方根误差分别降低0.0026、0.0091、0.001,表明所提方法能够有效去除近红外光谱中的噪声信息,并提高去噪的评估指标。

During the near infrared spectrum processing,the measured near infrared spectrum not only has the near-infrared characteristic spectrum of the tested sample,but also contains some random noise.The noise will affect the accuracy of the subsequent spectrum analysis.In order to improve the accuracy of the near infrared spectrum,denoising processing for near infrared spectrum is required.Single lifting wavelet transform denoising,wavelet transform denoising and Savitzky-Golay filter are difficult to obtain better denoising effect.Therefore,an improved method is proposed,which combines lifting wavelet transform and Savitzky-Golay filter for near infrared spectrum denoising.Meanwhile,the results are compared with those of single denoising methods by simulating and evaluating the denoising effect.The experimental results show that the signal-to-noise ratio of the new method is increased by 0.3364,1.1074,and 0.1287 respectively,and the root mean square error is reduced by 0.0026,0.0091,and 0.0010 respectively compared with corresponding values of wavelet transform denoising,lifting wavelet transform denoising and Savitzky-Golay filter.It indicates that the proposed method can effectively remove the noise in the near infrared spectrum and improve the evaluation index of noise denoising.