Automatic velocity picking based on deep learning

Zhang, Hao; Zhu, Peimin; Gu, Yuan; Li, Xiaozhang

doi:10.1190/segam2019-3215633.1

Cited by 25 publications

(11 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, these methods are only suitable for velocity picking under simple geological conditions and their degree of automation is low. To solve these problems, Zhang et al (2019) proposed a deep-learning method that uses a long short-term memory network to achieve velocity picking, thereby having high picking accuracy and degree of automation. Velocity-picking methods based on neural networks require sufficient labels for training the networks.…”

Section: Introductionmentioning

confidence: 99%

Automatic velocity picking with restricted weighted k-means clustering using prior information

Xie

Xu²,

Yang³

et al. 2023

Front. Earth Sci.

View full text Add to dashboard Cite

Automatic picking of seismic velocity can be performed using k-means clustering. In simple k-means clustering, the number of clusters needs to be predetermined, while the picking result is affected by the initial value of each cluster center. In this study, we present an unsupervised weighted k-means clustering velocity-picking method that picks the centers of the energy clusters instead of the geometric centers of the clusters. This method works on the semblance velocity spectrum and requires an initial velocity function and three user-defined thresholds to limit the search area. The number of cluster centers and their initial times are obtained according to a rectangular signal resulting from the three thresholds, while the initial velocities of the cluster centers can be subsequently obtained using their initial times and the initial velocity function. Inaccurate selection of thresholds may merge two clusters wrongly, in which case only a stronger event is selected. In the weighted k-means clustering algorithm, weights are calculated by using the amplitudes of the velocity points. Meanwhile, points far from the center are gradually removed to ensure that each cluster center coincides with the respective energy cluster center. We also propose a method for ignoring non-primary velocities, such as multiples, by removing points that create sudden changes in the slope of the reference velocity beyond a user-defined limit. The processing of the model and real data show that the proposed seismic velocity-picking method has high efficiency and picking accuracy.

show abstract

Section: Introductionmentioning

confidence: 99%

Automatic velocity picking with restricted weighted k-means clustering using prior information

Xie

Xu²,

Yang³

et al. 2023

Front. Earth Sci.

View full text Add to dashboard Cite

show abstract

“…Since both algorithms recognize NMO velocities in CMP gathers, the SNR of CMP gathers is required, which implies the picking results are no longer optimal for unclear trajectories. Zhang et al [18] employed You Only Look Once (YOLO), which is a state-of-the-art algorithm for object detection, to detect the energy blobs on the velocity spectrum. Additionally, long short-term memory (LSTM) network, a neural network for processing sequence data, was used for optimizing the final combination of the YOLO outputs.…”

Section: Introductionmentioning

confidence: 99%

Automatic Velocity Picking Using a Multi-Information Fusion Deep Semantic Segmentation Network

Wang,

Zhang,

Zhao

et al. 2022

Preprint

View full text Add to dashboard Cite

Velocity picking, a critical step in seismic data processing, has been studied for decades. Although manual picking can produce accurate normal moveout (NMO) velocities from the velocity spectra of prestack gathers, it is time-consuming and becomes infeasible with the emergence of large amount of seismic data. Numerous automatic velocity picking methods have thus been developed. In recent years, deep learning (DL) methods have produced good results on the seismic data with medium and high signal-to-noise ratios (SNR). Unfortunately, it still lacks a picking method to automatically generate accurate velocities in the situations of low SNR. In this paper, we propose a multiinformation fusion network (MIFN) to estimate stacking velocity from the fusion information of velocity spectra and stack gather segments (SGS). In particular, we transform the velocity picking problem into a semantic segmentation problem based on the velocity spectrum images. Meanwhile, the information provided by SGS is used as a prior in the network to assist segmentation. The experimental results on two field datasets show that the picking results of MIFN are stable and accurate for the scenarios with medium and high SNR, and it also performs well in low SNR scenarios.

show abstract

“…Second, the pickup problem is regarded as either an object detection problem, a semantic segmentation task or a time series prediction problem. Zhang et al [5] combined You Only Look Once (YOLO) and Long Short-Term Memory (LSTM) to build a discriminant model and Xu et al [6] built a regression model between normal moveout (NMO) correction time and velocity estimation losses by using a convolutional neural network (CNN) to find the energy peaks. Ferreira et al [7] simultaneously combined the information of both the velocity spectrum and the CMP gather for velocity analysis.…”

Section: Introductionmentioning

confidence: 99%

“…Even in different geological conditions, the chosen objective functions are different. For the supervised learning approaches, the deep learning models [5] [6] [7] [8] can achieve high picking accuracy. But as far as the current pickup model is concerned, its training tag velocity spectrum energy peak is a huge project and the trained model is difficult to generalize to other geological conditions.…”

Section: Introductionmentioning

confidence: 99%

Automatic Velocity Picking Using Unsupervised Ensemble Learning

Wang¹,

Zhang²,

Zhao³

et al. 2022

Preprint

View full text Add to dashboard Cite

In seismic data processing, accurate and efficient automatic velocity picking algorithms can significantly accelerate the processing, and the main branch is to use velocity spectra for velocity pickup. Recently, machine learning algorithms have been widely used in automatic spectrum picking. Even though deep learning methods can address the problem well in supervised cases, they are often accompanied by expensive computational costs and low interpretability. On the contrast, unsupervised learning methods based on the physical knowledge have great potential to efficiently resolve the task. In this paper, we propose an unsupervised ensemble learning (UEL) method to pick the root mean square (RMS) velocities on the spectrum. In particular, UEL utilizes the information of nearby velocity spectra and the nearest seed velocity curve to assist the selection of effective and reasonable velocity points. To increase the coherence of energy peaks, an information gain method is developed by local normalization. In addition, we designed the attention scale-space filter (ASSF) clustering method to incorporate the coherence information into the picking process. Experiments on three datasets demonstrate that compared to traditional clustering methods, UEL can recognize energy clusters better, especially with smaller blobs. Moreover, the injection of nearby spectra and interval velocity constraint in UEL significantly improves the robustness and accuracy of picking results.

show abstract

Automatic velocity picking based on deep learning

Cited by 25 publications

References 12 publications

Automatic velocity picking with restricted weighted k-means clustering using prior information

Automatic velocity picking with restricted weighted k-means clustering using prior information

Automatic Velocity Picking Using a Multi-Information Fusion Deep Semantic Segmentation Network

Automatic Velocity Picking Using Unsupervised Ensemble Learning

Contact Info

Product

Resources

About