Airborne electromagnetic data denoising based on dictionary learning  

基于字典学习的航空电磁数据去噪技术研究

作　　者：Xue Shu-yang Yin Chang-chun Su Yang Liu Yun-he Wang Yong Liu Cai-hua Xiong Bin Sun Huai-feng 薛舒杨;殷长春;苏扬;刘云鹤;王勇;刘才华;熊彬;孙怀凤(吉林大学地球探测科学与技术学院,吉林长春130026;吉林大学建设工程学院,吉林长春130026;中陕核工业集团二一四大队有限公司,西安710100;桂林理工大学地球科学学院,广西桂林541006;山东大学岩土与结构工程研究中心,山东济南250061)

机构地区：[1]College of Geo-Exploration Sciences and Technology,Jilin University,Changchun 130026,China [2]Construction Engineering College,Jilin University,Changchun 130026,China [3]Sino Shaanxi Nuclear Industry Group 214 Brigade Co.,Ltd,Xi'an 710100,China [4]College of Earth Sciences,Guilin University of Technology,Guilin 541006,China [5]Geotechnical and Structural Engineering Research Center,Shandong University,Jinan 250061,China.

出　　处：《Applied Geophysics》2020年第2期306-313,317,共9页应用地球物理（英文版）

基　　金：financially supported the Strategic Priority Research Program of the Chinese Academy of Sciences (No. XDA14020102);the National Natural Science Foundation of China (Nos. 41774125,41530320 and 41804098);the Key National Research Project of China (Nos. 2016YFC0303100,2017YFC0601900)。

摘　　要：Time-domain airborne electromagnetic(AEM)data are frequently subject to interference from various types of noise,which can reduce the data quality and affect data inversion and interpretation.Traditional denoising methods primarily deal with data directly,without analyzing the data in detail;thus,the results are not always satisfactory.In this paper,we propose a method based on dictionary learning for EM data denoising.This method uses dictionary learning to perform feature analysis and to extract and reconstruct the true signal.In the process of dictionary learning,the random noise is fi ltered out as residuals.To verify the eff ectiveness of this dictionary learning approach for denoising,we use a fi xed overcomplete discrete cosine transform(ODCT)dictionary algorithm,the method-of-optimal-directions(MOD)dictionary learning algorithm,and the K-singular value decomposition(K-SVD)dictionary learning algorithm to denoise decay curves at single points and to denoise profi le data for diff erent time channels in time-domain AEM.The results show obvious diff erences among the three dictionaries for denoising AEM data,with the K-SVD dictionary achieving the best performance.时间域航空电磁(AEM)勘探在测量过程中受各种噪声干扰,导致测量数据失真,影响了反演结果精度。传统的去噪方法大多针对数据本身进行加工,没有对数据的特征进行深入分析,所以去噪效果不理想。本文提出一种基于字典学习的航空电磁数据去噪方法。该方法是通过字典学习对含噪信号进行特征分析和提取,然后对信号进行重构。在字典学习过程中,把随机噪声作为残差过滤掉,达到去噪效果。为了验证所提去噪方法的有效性,我们将固定字典过完备离散余弦变换(Overcomplete discrete cosine transform,简称ODCT)、最优方向法(Method of optimal directions,简称MOD)字典学习算法和K-奇异值分解(K-singular value decomposition,简称K-SVD)字典学习算法用于时间域航空电磁单点衰减曲线和测线剖面上不同时间道数据去噪,结果表明三种字典学习方法对航空电磁数据去噪效果有明显差异,以K-SVD字典学习方法效果最佳。

关 键 词：Time-domain AEM data processing DENOISING dictionary learning sparse representation 

分 类 号：TN911.4[电子电信—通信与信息系统] P631.325[电子电信—信息与通信工程]