基于改进双树复小波的光谱去噪算法研究  

Research on spectrum denoising based on improved dual-tree complex wavelet transform

作　　者：张立国[1] 胡永涛[1] 张淑清[1] 李军锋[1] 吴迪[1] 姜万录[1] Zhang Liguo Hu Yongtao Zhang Shuqing Li Junfeng Wu Di Jiang Wanlu(Institute of Electrical Engineering, Yanshan University, Qinhuangdao 066004, Chin)

机构地区：[1]燕山大学电气工程学院,秦皇岛066004

出　　处：《仪器仪表学报》2016年第9期2061-2067,共7页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(51475405;61077071);河北省自然科学基金(F2015203413;F2015203392)项目资助

摘　　要：为了消除可见光近红外光谱噪声,提高利用光谱曲线进行信息提取的精度,提出一种改进双树复小波变换(DTCWT)的后验估计及广义形态滤波的光谱去噪方法。首先对带噪信号进行双树复小波分解,将信号的高频部分和低频部分进行分离。然后分别采用最大后验(MAP)估计算法和广义形态学滤波(GMF)对高频系数和低频系数进行去噪处理。最后对去噪后的高频系数和低频系数进行DTCWT反变换,得到去噪光谱。对USGS光谱库中的植被光谱以及铁铝榴石光谱进行实验,结果表明该方法易于实现,去噪效果理想,是一种很好的可见光近红外光谱去噪方法。In order to remove the noise in the visible and near infrared spectra and to improve the accuracy of information extraction with the spectrum, an improved dual-tree complex wavelet transform （DTCWT） denoising method based on maximum a posteriori （MAP） estimation and generalized morphological filter（GMF） is presented. Firstly, the noisy signal is decomposed into high frequency and low frequency parts with the DTCWT. Then, MAP estimation and GMF are adopted for high frequency and low frequency denoising respectively. Finally, the denoised spectrum is obtained by reconstituting denoised high frequency and low frequency parts. Vegetables and almandineis spectra from the USGS spectral library are used in experiments, and the results show that the proposed method is ideal for denoising, which is easier to implement. A good denoising method is provided for visible and near infrared spectra.

关 键 词：可见光近红外光谱 双树复小波 最大后验估计 广义形态滤波 去噪 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置] TH183.3[自动化与计算机技术—控制科学与工程]