Desert low-frequency noise suppression by using adaptive DnCNNs based on the determination of high-order statistic

Dong, Xintong; Li, Y; Yang, Banghua

doi:10.1093/gji/ggz363

Cited by 76 publications

(14 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(19). Rectifier units help to find better minima than other activation functions during training [44]. In addition, the receptive field size of the network is (2d + 1) × (2d + 1), that is, 35 × 35.…”

Section: Appendix a Rbmmentioning

confidence: 99%

Spectra-GANs: A New Automated Denoising Method for Low-S/N Stellar Spectra

Pan

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Numerous spectra can be obtained from sky surveys such as the Sloan Digital Sky Survey and the Large Sky Area Multi-Object Fibre Spectroscopic Telescope. However, a considerable fraction of such spectra, which are also valuable for astronomical research, are of low quality, possessing characteristics such as low signal-to-noise ratio (low-S/N). Principal component analysis is widely used to process these low-S/N spectra, but it is not efficient enough to describe the non-linear properties within the spectra. Wavelets are often used to denoise the low-S/N spectra. However, as is well known, the most optimal wavelet basis for each type of spectra needs to be determined; therefore, wavelet analysis is very difficult to use in practice. Restricted Boltzmann machine is a non-linear algorithm that performs poorly when applied to low-S/N spectra. Denoising Convolutional Neural Networks (DnCNN) is a promising denoiser, however, its performance is unsatisfactory due to the lack of suitable noise model. To better exploit the spectra with low-S/N, we propose a new method that can be used to obtain better denoised spectra when compared to those obtained using other methods. A new method called the Spectra Generative Adversarial Nets (Spectra-GANs) is introduced. Spectra-GANs is simply a feedforward neural network that learns the difference between the input vector and the target by minimizing the loss function. It can be used in spectral denoising. The performance of Spectra-GANs is better than those of other methods with regard to denoising the spectra, especially with regard to extremely low-S/N spectral processing. Thus, Spectra-GANs proposed herein is a suitable alternative to previously used methods in spectral denoising.

show abstract

Section: Appendix a Rbmmentioning

confidence: 99%

Spectra-GANs: A New Automated Denoising Method for Low-S/N Stellar Spectra

Pan

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Yu et al . (2019) gave a detailed overview on using CNN models for noise suppression in seismic data (also see Dong et al ., 2019, for low‐frequency de‐noising). Sun et al .…”

Section: Introductionmentioning

confidence: 99%

“…Perol et al (2018) used the feed-forward CNN for seismological record classification and approximate location. Yu et al (2019) gave a detailed overview on using CNN models for noise suppression in seismic data (also see Dong et al, 2019, for low-frequency de-noising). Sun et al (2018) performed an analysis of seismic gathers to identify if the shingling effect was present in the first-arrival of waveforms.…”

Section: Introductionmentioning

confidence: 99%

Convolution neural network application for first‐break picking for land seismic data

Loginov

Duchkov

Litvichenko

et al. 2022

Geophysical Prospecting

View full text Add to dashboard Cite

An automatic and robust algorithm for the first-break picking is necessary to build the near-surface velocity model. We propose the algorithm based on a convolution neural network. The introduced first-break picking strategy and neural network architecture are suited for processing large volumes of seismic exploration data with reasonable computational resources. To develop an optimal neural network topology and architecture, extensive testing was performed. We compared several architectures of neural networks, including one-and two-dimensional approaches. Our tests justify that the one-dimensional approach (trace-by-trace processing) provides the most reliable results in the case of first-break travel-time variations typical of complicated near-surface structures. This study demonstrates that the four-layered neural network trained on 5,000 traces is enough for robust first-break picking. The algorithm is evaluated on two land-acquisition field datasets from West Siberia with a total used size of about 7 million traces. The first dataset is used for training, and the second one is used only for testing. For both datasets, the error between the original and the predicted first breaks is not more than three samples for 95% of traces. The final evaluation is done by a comparison of seismic stacks to prove the benefits of the approach and its robustness for offsets over 600 m. Finally, the influence of choosing the locations for the training dataset is discussed, and a strategy for using the proposed approach in production work is introduced.

show abstract

“…For the method used in [58], the learning goal in the training phase is an effective signal, which greatly increased the amount of computation of the network, thereby affected the training efficiency of the network. Dong et al [59] uses adaptive DnCNNs to denoise noisy desert seismic data, which effectively improves the signal-tonoise ratio of low-frequency noise. Chen et al [44] based on the concept of self-encoder in unsupervised learning, adaptively learns seismic signals from noise, and achieves random noise suppression of seismic data.…”

Section: Introductionmentioning

confidence: 99%

Random Noise Attenuation Based on Residual Convolutional Neural Network in Seismic Datasets

et al. 2020

View full text Add to dashboard Cite

Seismic random noise attenuation is a key step in seismic data processing. The random seismic data recorded by the detector tends to have strong noise, and this noisy seismic ratio can be seen as a low signal-to-noise ratio (SNR). Low SNR data can seriously affect the subsequent processing of seismic data, such as migration and imaging. Therefore, it is crucial to eliminate random noise in seismic data. In this paper, we aimed to improve the SNR of seismic data, and proposed an intelligent convolutional neural network noise reduction framework to adaptively capture seismic signals from seismic data with noise. The eponential linear unit (ELU) activation function and the Adam optimization algorithm were used to train the network, which increased the effective information extraction of the network in the negative interval. In order to speed up network training, we added residual learning and batch normalization methods to the network. In addition, three datasets were used to train and test the network. The experimental results show that the method proposed in this paper is better than feed-forward denoising convolutional neural networks (DnCNNs) and other contrast methods in denoising performance. More importantly, a well-trained network not only preserves weak features in learning, but also removes spatially random noise. First, the proposed method is fully trained to extract random noise from the training data, then we retain the data features learned in the training, and estimate the waveform characteristics in the test set by reconstructing the recorded seismic data. Secondly, the characteristics of seismic data in the field record are quite different from those of the training set. However, the proposed adaptive denoising framework can still capture the connection between prediction and reality from the difference. The processing results of theoretical pure record, common-shot-point record, and field data showed great potential in random noise attenuation applications.INDEX TERMS Deep learning, residual convolutional neural network, seismic data denoising, visualization.

show abstract

Desert low-frequency noise suppression by using adaptive DnCNNs based on the determination of high-order statistic

Cited by 76 publications

References 55 publications

Spectra-GANs: A New Automated Denoising Method for Low-S/N Stellar Spectra

Spectra-GANs: A New Automated Denoising Method for Low-S/N Stellar Spectra

Convolution neural network application for first‐break picking for land seismic data

Random Noise Attenuation Based on Residual Convolutional Neural Network in Seismic Datasets

Contact Info

Product

Resources

About