基于稀疏表示和压缩感知的地震信号处理及应用研究
发布时间：2018-10-16 10:30
【摘要】：随着地震勘探逐步走入复杂地形勘探的环境,因而所获得的地震信号的信噪比较低。同时这些复杂地形中的地震信号,往往存在着较为严重的信号缺失和废道,废炮,这为地震信号的处理提出了新的要求和挑战。而由于地震信号在频域的稀疏性,其可以使用较少量的系数来表征整个地震信号,同时,各种使得地震信号质量下降的污染都可以被视为对这种稀疏性的破坏。因而该性质可以被有效地应用其来对地震信号进行更加精确的处理,从而提高地震剖面的质量。本文围绕地震信号的去噪,地震信号的缺失道重建以及超分辨率重建这三个主要问题,从深入学习稀疏表示和压缩感知的理论框架出发,重点研究了稀疏表示和压缩感知理论在地震信号处理流程中得典型应用。主要研究内容如下:对于地震去噪问题,本文在研究了Shearlet变换的理论与实现和阈值去噪的基础上,提出了运用各向异性扩散滤波器对子带进行处理的去噪方法,达到了同时在Shearlet域上进行了信号增强和噪声抑制的目的。同时,本文对其进行了数值仿真与分析另外,本文针对地震数据的缺失道重构问题,在对地震缺失道的模型进行了分析的基础上,针对多种促稀疏重构方法进行的研究,应用与详细的结果分析。并针对这些方法在地震数据重构领域中的应用范围进行了对比和分析。继而本文设计了一种两步法进行地震数据的超分辨率重建的策略,依据不同的数据规模,选择不同的方法和稀疏表示来分别对时间分辨率和空间分辨率进行处理。并对理论地震数据以及实际地震数据进行了算法的仿真与测试,取得了较好的效果。
[Abstract]:The signal-to-noise ratio (SNR) of the obtained seismic signal is low with the seismic exploration gradually moving into the environment of complex terrain exploration. At the same time, the seismic signals in these complex terrain often have more serious signal missing, abandoned road, abandoned gun, which put forward new requirements and challenges for seismic signal processing. Due to the sparsity of seismic signal in frequency domain, a small number of coefficients can be used to characterize the whole seismic signal. At the same time, all kinds of pollution that make the quality of seismic signal decline can be regarded as damage to the sparsity. Therefore, this property can be effectively used to process seismic signals more accurately, thus improving the quality of seismic profiles. This paper focuses on the three main problems of seismic signal denoising, seismic signal missing channel reconstruction and super-resolution reconstruction, starting from the theoretical framework of sparse representation and compression perception. The typical application of sparse representation and compression sensing theory in seismic signal processing is studied. The main research contents are as follows: for the problem of seismic denoising, based on the study of the theory and implementation of Shearlet transform and threshold denoising, an anisotropic diffusion filter is proposed to deal with subband denoising. The aim of both signal enhancement and noise suppression in Shearlet domain is achieved. At the same time, numerical simulation and analysis are carried out. In addition, aiming at the problem of reconstruction of missing traces in seismic data, based on the analysis of the model of seismic missing traces, this paper studies several methods to promote sparse reconstruction. Application and detailed result analysis. The application scope of these methods in seismic data reconstruction is compared and analyzed. Then this paper designs a two-step method for super-resolution reconstruction of seismic data. According to different data scale, different methods and sparse representation are selected to process temporal resolution and spatial resolution respectively. The theoretical seismic data and the actual seismic data are simulated and tested, and good results are obtained.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P631.4;TN911.7
本文编号：2274076
[Abstract]:The signal-to-noise ratio (SNR) of the obtained seismic signal is low with the seismic exploration gradually moving into the environment of complex terrain exploration. At the same time, the seismic signals in these complex terrain often have more serious signal missing, abandoned road, abandoned gun, which put forward new requirements and challenges for seismic signal processing. Due to the sparsity of seismic signal in frequency domain, a small number of coefficients can be used to characterize the whole seismic signal. At the same time, all kinds of pollution that make the quality of seismic signal decline can be regarded as damage to the sparsity. Therefore, this property can be effectively used to process seismic signals more accurately, thus improving the quality of seismic profiles. This paper focuses on the three main problems of seismic signal denoising, seismic signal missing channel reconstruction and super-resolution reconstruction, starting from the theoretical framework of sparse representation and compression perception. The typical application of sparse representation and compression sensing theory in seismic signal processing is studied. The main research contents are as follows: for the problem of seismic denoising, based on the study of the theory and implementation of Shearlet transform and threshold denoising, an anisotropic diffusion filter is proposed to deal with subband denoising. The aim of both signal enhancement and noise suppression in Shearlet domain is achieved. At the same time, numerical simulation and analysis are carried out. In addition, aiming at the problem of reconstruction of missing traces in seismic data, based on the analysis of the model of seismic missing traces, this paper studies several methods to promote sparse reconstruction. Application and detailed result analysis. The application scope of these methods in seismic data reconstruction is compared and analyzed. Then this paper designs a two-step method for super-resolution reconstruction of seismic data. According to different data scale, different methods and sparse representation are selected to process temporal resolution and spatial resolution respectively. The theoretical seismic data and the actual seismic data are simulated and tested, and good results are obtained.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P631.4;TN911.7
【参考文献】
相关硕士学位论文 前1条
1 魏芳;基于多尺度几何分析的红外弱小目标检测方法研究[D];电子科技大学;2012年
，本文编号：2274076
本文链接：https://www.wllwen.com/kejilunwen/wltx/2274076.html
