The proposed blood cell target detection algorithm based on YOLOv5 addresses the issue of low average accuracy and serious miss detection due to small blood cells and serious cell adhesion in blood cell detection by target detection algorithms. By adding the CBAM (Convolutional Block Attention Module) to the YOLOv5 framework's backbone network and the BIFPN (bidirectional feature pyramid network) to the neck network, the algorithm improves the model's ability to extract features. The experimental results show that the average accuracy (mAP) of the improved YOLOv5 blood cell target detection algorithm is 89.9%, representing a increase over the native YOLOv5s type, and the recall rate and accuracy rate are also increased by 3.2% and 4.2%, respectively. This meets the requirements of the actual scene for blood cell detection.

Seismic data are always in low-resolution due to the limitations of seismic acquisition and processing technology, which bring challenges to subsequent seismic interpretation. Deep learning has been successfully applied to the seismic data super-resolution, all methods tend to directly learn the mapping relationship between low-resolution seismic images and super-resolution seismic images through some complex convolutional neural networks. But blindly increasing the depth of the network brings limited improvement to the super-resolution work. We propose a novel geophysical prior guided framework for seismic data super-resolution to solve this problem. Specifically, we select fault prior to guide the training of deep learning model: we use knowledge distillation technology to progressively propagate the fault prior from the teacher network (trained with the low-resolution synthetic seismic data/high-resolution fault prior and high-resolution synthetic seismic data pairs) to the student net-work (trained with the low-resolution synthetic seismic data and high-resolution synthetic seismic data pairs). To better propagate fault priors, we use feature space loss and soft ground truth loss in student network training. Finally, we use the trained student network to complete the super-resolution of synthetic validation seismic data and real seismic data. In addition, in our super-resolution framework, we directly process 3D seismic data instead of 2D seismic images, which further improves the effects of super-resolution and the subsequent seismic interpretation. Compared with the state-of-the-art seismic super-resolution method, the experimental results show that the super-resolution results of our method can depict the faults more clearly.

Fault detection is a key step in seismic structure interpretation. Current research has achieved good results in fault detection by using synthetic training data to train deep-learning models. However, there is an inevitable difference in frequency bandwidth between synthetic training data and real seismic data, which makes it difficult for deep-learning models to obtain ideal fault detection results on real seismic data. To solve this problem, the feature pyramid network (FPN) is introduced to obtain multiscale deep-learning features, which can reduce the impact of seismic data frequency bandwidth differences on fault detection. Then, we apply the multiscale wavelet transform to extract multiscale frequency spectral features of the seismic data and combine them with the multiscale deep-learning features through concatenation operation. Furthermore, the seismic data is decomposed into signals with different frequency bands through the wavelet transform, and we use the energy of these signals as the network weights of multiscale mixed features to further improve the frequency adaptability of our method. Based on these works, we not only improve the fault detection effect in a specific work area but also improve the generalization ability of the deep-learning model in different work areas, thus further promoting the application of deep learning in actual production. Compared with the fault detection results by the traditional deep-learning model U-Net and the traditional FPN on multiple real seismic data and synthetic seismic data, experimental results demonstrate the effectiveness of our method.

Fault surface extraction plays a vital role in structural interpretation and structural modeling, which can enhance our understanding of geologic structures. Significant effort has been invested in fault surface extraction in the past few years. These contributions can be roughly divided into two types: one constructs fault sticks or fault patches and links them to form fault surfaces, and the other constructs a complete fault surface by roughly determining the location of a fault. The former is likely to extract incomplete fault surfaces in areas with low data quality or complex fault structures. The latter has problems dealing with faults in the case of complex fault structures, such as intersecting faults. To effectively address these problems, we have developed a novel fault extraction workflow from a global perspective, which can extract fault surfaces on the premise of obtaining the fault distribution of the entire data set. First, we introduce a multilayer complex network to characterize the overall distribution of faults, in which nodes represent faults, node attributes cover information, such as the location of the faults, and edge attributes give the relationship between faults. Then, for each fault represented by the node, we develop a fault surface extraction method based on computational topology, which can guarantee the completeness of the fault surfaces. We apply our method to synthetic data and field seismic data to evaluate the method quantitatively and qualitatively, respectively. The results indicate that the proposed approach extracts the fault effectively, ensures the completeness of the fault surface, and has high performance in areas with complex faults. Thus, the approach can work well in complex situations, such as intersecting faults.

Seismic attributes are critical in understanding geologic factors, such as sand body configuration, lithology, and porosity. However, existing attributes typically reflect the combined response of multiple geologic factors. The interplay between these factors can obscure the features of the target factor, posing a challenge to its direct seismic characterization, particularly when the factor is subtle. To address this, we develop an innovative neural network designed to disentangle and characterize the individual geologic factors within seismic data. Our approach divides the geologic information in the seismic data into two categories: a single geologic factor of interest and an aggregate of all other information. A novel feature-swapping mechanism within our network facilitates the disentanglement of these two categories, providing an interpretable representation. We use a triplet loss function to differentiate data samples with similar waveforms but varying subtle geologic details, thus enhancing the extraction of distinct features. In addition, our network uses a cotraining strategy to integrate the synthetic and actual field data during the training process. This strategy helps mitigate the potential performance degradation arising from the discrepancies between simulated and actual field data. We apply our method to synthetic data experiments and field data from two geologically distinct areas. Current results indicate that our method surpasses traditional approaches, such as a deep autoencoder and a convolutional neural network classifier, in extracting seismic attributes with more explicit geophysical implications.

Seismic 3D full-horizon tracking is a fundamental and crucial step in sequence analysis and reservoir modeling. Existing automatic full-horizon tracking approaches lack effective methods for representing the stratigraphic sequence relationships in seismic data. However, the inability to represent the stratigraphic sequence relationships fully and accurately makes it challenging to address discontinuous areas affected by faults and unconformities. To address this issue, we develop a knowledge graph representing the stratigraphic sequence relationship, which enables the simultaneous extraction of all horizon surfaces once the stratigraphic distribution of the seismic data is obtained. This method first generates horizon patches and calculates the fault attributes, followed by the construction of an initial knowledge graph that characterizes the overall distribution of horizon patches and faults. The initial knowledge graph comprises nodes and edges. The nodes represent horizon patches, and their attributes cover the geographical location information of the patches and faults. Simultaneously, the edges represent the relationship between the horizon patches, including the stratigraphic sequence relationship, and their attributes illustrate the potential for connecting these patches. Furthermore, we introduce a multilayer knowledge graph based on the point-set topology to fuse the nodes. This allows for the continuous merging of the horizon patches to obtain horizon surfaces across discontinuities with the constraints of fault attributes and stratigraphic sequence relationships in 3D space. Synthetic and field examples demonstrate that our approach can effectively represent stratigraphic sequence relationships and accurately track horizons located in discontinuous areas with faults and unconformities.