Adaptive threshold shearlet transform for surface microseismic data denoising
Adaptive threshold shearlet transform for surface microseismic data denoising
复制标题
地面微震数据去噪的自适应阈值剪切波变换
DOI：
10.1016/j.jappgeo.2018.03.019
复制
发表时间：
2018-06-01
影响因子：
2
通讯作者：
Zhu, Dan
中科院分区：
文献类型：
--
作者：
Tang, Na;Zhao, Xian;Zhu, Dan
随机噪声的压制在微地震资料处理中占有重要地位。微地震资料中经常会受到强随机噪声的干扰，这将直接影响微地震事件的识别和定位。Shearlet变换是一种新的多尺度变换，可以有效地处理低震级的微地震数据。在剪切波域中，由于有效信号和随机噪声的分布不同，剪切波系数可以通过阈值进行收缩。因此，阈值在抑制随机噪声中起着至关重要的作用。传统的阈值去噪算法通常采用相同的阈值对所有系数进行去噪处理，导致噪声抑制效率低下或有效信号丢失。为了解决上述问题，本文提出了一种自适应阈值剪切波变换（ATST）的地表微地震数据去噪方法。在新算法中，我们首先计算每个方向子带的基本阈值。在每个方向子带中，根据每个子带系数及其相邻子带系数确定调整因子，以自适应地调整不同剪切波系数的基本阈值。最后，我们采用自适应阈值来处理不同的剪切波系数。模拟记录和现场数据的去噪实验结果表明，该方法在抑制随机噪声和保留有效信号方面优于传统的剪切波去噪方法。(C)2018由Elsevier B.V.出版
Random noise suppression plays an important role in microseismic data processing. The microseismic data is often corrupted by strong random noise, which would directly influence identification and location of microseismic events. Shearlet transform is a new multiscale transform, which can effectively process the low magnitude of microseismic data. In shearlet domain, due to different distributions of valid signals and random noise, shearlet coefficients can be shrunk by threshold. Therefore, threshold is vital in suppressing random noise. The conventional threshold denoising algorithms usually use the same threshold to process all coefficients, which causes noise suppression inefficiency or valid signals loss. In order to solve above problems, we propose the adaptive threshold shearlet transform (ATST) for surface microseismic data denoising. In the new algorithm, we calculate the fundamental threshold for each direction subband firstly. In each direction subband, the adjustment factor is obtained according to each subband coefficient and its neighboring coefficients, in order to adaptively regulate the fundamental threshold for different shearlet coefficients. Finally we apply the adaptive threshold to deal with different shearlet coefficients. The experimental denoising results of synthetic records and field data illustrate that the proposed method exhibits better performance in suppressing random noise and preserving valid signal than the conventional shearlet denoising method. (C) 2018 Published by Elsevier B.V.