Real-time image super-resolution reconstruction method, system and device for resistivity logging while drilling based on generative adversarial network and medium
The image super-resolution reconstruction method using generative adversarial networks solves the problem of insufficient real-time data resolution in logging-while-drilling resistivity, thereby improving image quality and risk management capabilities during the drilling process.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XI'AN PETROLEUM UNIVERSITY
- Filing Date
- 2024-09-12
- Publication Date
- 2026-05-19
AI Technical Summary
In existing technologies for logging resistivity while drilling, there is a significant resolution gap between real-time data and stored data, which affects decision support and risk management during the drilling process.
A generative adversarial network-based approach is adopted to construct a generator and a discriminator, extract image features using convolutional layers, SE-RRDB modules, and upsampling modules, and optimize training parameters by combining a loss function to achieve super-resolution reconstruction of low-resolution images.
It improves the resolution and clarity of real-time images, optimizes decision support and risk management capabilities during the drilling process, and provides higher quality data support.
Smart Images
Figure CN119228647B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of oil well exploration technology, specifically to a method, system, equipment, and medium for real-time super-resolution reconstruction of resistivity logging images during drilling based on generative adversarial networks. Background Technology
[0002] Imaging logging is an indispensable tool in modern geological exploration and oil and gas development. It uses instruments to scan the wellbore, acquiring and processing various signal data such as resistivity, acoustic waves, and nuclear magnetic resonance (NMR) to generate high-resolution two-dimensional or three-dimensional images. This allows for detailed analysis of formation lithology, structural features, fracture distribution, and porosity, providing precise data support for geological research and reservoir characterization. Among the images generated by imaging logging, resistivity images are highly sensitive to changes in the electrical properties around the wellbore, thus offering the highest resolution. With resistivity images, even very small features, such as natural or induced fractures, can be clearly identified.
[0003] Initially, logging images were obtained through scanning imaging using cable-based instruments. Cable logging involved pressing sensors against the wellbore wall, allowing them to slide as the tool passed through the wellbore. However, this method was prone to sensor wear during drilling. Logging while drilling (LOD) uses a rotating drill string as a sensor platform, moving the sensors along a helical path, thus reducing wear. By transmitting sensor measurements to the surface in real time, a range of problems that cable logging techniques could not address were effectively solved. Key areas of focus include wellbore integrity, geosteering, and completion optimization. High-resolution real-time imaging significantly enhances the effectiveness of these applications by identifying wellbore events and formation features. High-resolution real-time imaging improves wellbore stability management by analyzing drilling-induced fractures, wellbore fracturing, and other issues. Geosteering utilizes high-resolution real-time data to precisely adjust the drilling direction, ensuring the wellbore is optimally positioned within the target formation. High-resolution data enables more accurate real-time bedding assessment and structural analysis, optimizing the drilling path and reducing drilling risks. In the completion phase, key geological structures that can affect communication or cementation quality between production or injection zones allow for faster completion decisions. In Trevizan WA, de Jesus C M. Application of GAN to Resolution Enhancement of LWD Real-Time Image Logs to Support Decision Making[J]. Petrophysics-The SPWLA Journal of Formation Evaluation and Reservoir Description, 2023, 64(06):890-899., Willian et al. used a generative adversarial network-based model to perform super-resolution reconstruction of low-resolution real-time images, obtaining higher-quality super-resolution images to facilitate timely identification of geological and geomechanical structures and support well completion decisions. Although they achieved good results, they did not measure the effects of different hyperparameter combinations on the super-resolution reconstruction of low-resolution images. Summary of the Invention
[0004] To address the shortcomings of the existing technology, the present invention aims to provide a method, system, device, and medium for real-time super-resolution reconstruction of resistivity logging images during drilling based on generative adversarial networks; constructing a dataset of high-resolution stored data images and low-resolution real-time data images; using a deep learning algorithm to train the constructed paired image datasets; and performing super-resolution reconstruction of the low-resolution images after training to approximate the resolution of the stored data. This not only compensates for the deficiencies of real-time data but also optimizes decision support and improves risk management capabilities during the drilling process.
[0005] To achieve the above objectives, the present invention adopts the following technical solution:
[0006] A method for super-resolution reconstruction of real-time images from drilling resistivity logging based on generative adversarial networks, specifically including the following steps:
[0007] Step 1: By reading the logging resistivity data from different regions, the logging resistivity data is sampled at equal intervals and then imaged to obtain the corresponding low-resolution image; the logging resistivity data is mapped to a pixel range of 0-255, and imaged according to pixels to obtain logging images as high-resolution images for storage; the high-resolution image and the corresponding low-resolution image form a paired image dataset; the paired image dataset is augmented to obtain an augmented image dataset;
[0008] Step 2: Based on the expanded image dataset obtained in Step 1, build a generative adversarial network model;
[0009] Step 3: Use the expanded image dataset obtained in Step 1 to train and optimize the generative adversarial network model in Step 2, and save the optimal generative adversarial network model.
[0010] Step 4: Obtain real-time logging data of resistivity in the target area while drilling. Map the real-time logging data of resistivity in the target area to a pixel range of 0-255. After imaging according to the pixels, input it into the generative adversarial network model trained in Step 3 to obtain the super-resolution real-time image. Compare it with the high-resolution image stored in the logging data while drilling to measure the super-resolution effect.
[0011] The specific method for building the generative adversarial network model in step 2 is as follows:
[0012] The generative adversarial network model consists of a generator and a discriminator. The generator takes the low-resolution image obtained in step 1 as input and includes convolutional layers, an SE-RRDB module, and an upsampling module. The generator extracts the basic features of the image through the convolutional layers; the SE-RRDB module extracts and fuses the basic features of the image; the upsampling module adjusts the number of channels in the convolutional layers and samples the fused image features to the target output size; and the low-resolution image is transformed into a super-resolution image output.
[0013] The discriminator uses a U-Net structure with spectral normalization. It evaluates the input high-resolution image and super-resolution image through multi-layer convolution and skip connections, determines whether the high-resolution image and super-resolution image are similar, and feeds the discrimination result back to the generator to guide the generator to generate more realistic images.
[0014] Set the loss function for the generative adversarial network model to adjust the network parameters of the generator and discriminator until the final generative adversarial network model is obtained. The loss function is:
[0015] L SR =c1L1+c p L p
[0016] L1 = ||I out -I gt ||1
[0017]
[0018] Among them, L SR Let L1 be the loss function, and c be the pixel loss function. adv To counteract the weighting, L p For the perceptual loss function, c p For perceived weights. I out It is an image generated by the generator, I gt It is a true high-resolution image; |||1 represents the L1 norm; N is the number of feature maps, φ i () is the feature map of the i-th layer of the pre-trained VGG network.
[0019] The process of step 3 is as follows:
[0020] The paired image dataset from step 1 is divided into a training set and a test set. The generative adversarial network (GAN) model is trained using the training set, and the test set data is imported into the trained GAN model for testing. Super-resolution reconstruction is performed on the low-resolution images. Based on the super-resolution reconstruction results, the hyperparameters of the GAN model are adjusted. The peak signal-to-noise ratio (PSNR) and structural similarity of the images are statistically analyzed to find the optimal hyperparameter settings. The network model corresponding to the optimal hyperparameters is the optimal GAN model, and the optimal GAN model is saved.
[0021] The hyperparameters include learning rate, batch size, and loss function weights.
[0022] The method for finding the optimal hyperparameter settings is as follows: use different combinations of learning rate and batch size to verify the super-resolution reconstruction effect of the generative adversarial network under different training parameters, and select the learning rate and batch size corresponding to the minimum average peak signal-to-noise ratio and structural similarity as the optimal parameters of the generative adversarial network model.
[0023] A real-time super-resolution reconstruction system for logging resistivity images while drilling, based on generative adversarial networks, includes:
[0024] The first acquisition module is used to acquire the drilling resistivity storage data from different regions, map the drilling resistivity storage data to a pixel range of 0-255, and image it into a high-resolution image according to the pixels. The drilling resistivity storage data is sampled at equal intervals and then imaged to obtain the corresponding low-resolution image; it is used in step 1 to realize the construction of the dataset.
[0025] The Generative Adversarial Network (GAN) model building module is used to build a GAN model, which includes a generator and a discriminator; that is, it is used in step 2 to build the GAN model.
[0026] The generator includes convolutional layers, an SE-RRDB module, and an upsampling module. The convolutional layers extract basic image features; the SE-RRDB module extracts and fuses basic image features; the upsampling module adjusts the number of channels in the convolutional layers and samples the fused image features to the target output size; this is used in step 2 to transform a low-resolution image into a super-resolution image. The SE-RRDB module includes three sequentially connected RDB sub-modules and one SENet sub-module; the SENet sub-module includes a sequentially connected global average pooling layer, a fully connected layer, and an activation function, which improves the performance of the convolutional neural network by introducing a channel attention mechanism; the RDB sub-module includes multiple sequentially connected convolutional layers, and the output of each convolutional layer is concatenated with the output of the previous convolutional layer.
[0027] The discriminator, consisting of multiple convolutional layers, incorporates skip connections between each convolutional layer to preserve low-level features, helping the discriminator better capture multi-scale information. In step 2, it is used to obtain discrimination results from the input high-resolution image and super-resolution image, thereby improving the final effect of super-resolution reconstruction.
[0028] The feedback module sets the loss function of the generative adversarial network model to adjust the network parameters of the generator and discriminator until the final generative adversarial network model is obtained.
[0029] The network optimization module is used in step 3 to divide the paired image dataset constructed in step 1 into a training set and a test set. The training set is used to train the generative adversarial network (GAN) model, and the test set data is imported into the trained GAN model for testing. Super-resolution reconstruction is performed on low-resolution images. Based on the super-resolution results, the hyperparameters of the GAN model are adjusted. The peak signal-to-noise ratio (PSNR) and structural similarity of the images are statistically analyzed to find the optimal hyperparameter settings. The GAN model corresponding to the optimal hyperparameters is the optimal super-resolution model, and the optimal super-resolution model is saved.
[0030] The second acquisition module is used in step 4 to acquire the logging-while-drilling resistivity data of the target area and scale it as a logging-while-drilling resistivity image, thereby realizing the acquisition of the real-time logging-while-drilling resistivity image of the target area and the stored logging-while-drilling resistivity image.
[0031] The super-resolution module is used to input the real-time logging data of the target resistivity while drilling into the super-resolution model to obtain a super-resolution image, which is used in step 4 to realize the super-resolution reconstruction of the real-time logging image of low-resolution resistivity while drilling.
[0032] The evaluation module is used in step 4 to compare the real-time image after super-resolution reconstruction with the processed and stored data to measure the super-resolution effect.
[0033] A real-time super-resolution image device for logging resistivity while drilling based on generative adversarial networks, comprising:
[0034] The system includes a memory and a processor. The memory stores a computer program that, when executed, performs super-resolution reconstruction of real-time images based on the generative adversarial network-based real-time image super-resolution reconstruction method for logging resistivity while drilling, as described in steps 1 to 4, thereby improving image detail information and enhancing image resolution and clarity.
[0035] A computer-readable storage medium storing a computer program provides the function of storing and executing an image processing program. When executed by a processor, the computer program can achieve efficient and fast real-time super-resolution reconstruction of real-time images of logging resistivity while drilling based on the generative adversarial network-based super-resolution reconstruction method described in steps 1 to 4.
[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0037] Due to the slow communication speed of mud pulse telemetry technology and the increasing data volume in measurement-while-drilling systems, the gap between real-time data and stored data has widened further. Therefore, this invention performs super-resolution reconstruction of real-time data to approximate the resolution of stored data. This not only compensates for the deficiencies of real-time data but also optimizes decision support and improves risk management capabilities during the drilling process.
[0038] This invention achieves more refined and effective real-time image super-resolution processing within generative adversarial networks by introducing an attention mechanism and optimizing training parameters. Compared to existing technologies, this invention significantly improves detail reconstruction, training stability, and practical application performance, providing higher-quality images and more reliable data support for practical applications such as logging-while-drilling resistivity.
[0039] This invention proposes a real-time image super-resolution reconstruction method based on generative adversarial networks (GANs). By introducing an attention mechanism into the generator, the improved generator can fuse more image information. Optimal training parameters are used to perform super-resolution reconstruction on real-time images, and the super-resolution effect is tested under different resolution conditions. This provides more reliable results for subsequent optimization of well completion stages using real-time images from logging-while-drilling resistivity. Attached Figure Description
[0040] Figure 1 This is a flowchart of the super-resolution reconstruction process of the present invention.
[0041] Figure 2 This is a schematic diagram of the overall network structure of the present invention.
[0042] Figure 3 This is a schematic diagram of the generator structure of the present invention.
[0043] Figure 4 This is a schematic diagram of the SE-RRDB submodule of the present invention.
[0044] Figure 5 This is a schematic diagram of the SENet submodule of the present invention.
[0045] Figure 6 This is a schematic diagram of the discriminator of the present invention.
[0046] Figure 7 This is a rendering of the super-resolution reconstruction. Detailed Implementation
[0047] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.
[0048] Please see Figure 1 A method for super-resolution reconstruction of real-time images from drilling resistivity logging based on generative adversarial networks includes the following steps:
[0049] Step 1. By reading the logging resistivity data stored in different regions, the logging resistivity data is sampled at equal intervals and then imaged to obtain the corresponding low-resolution image; the logging resistivity data is mapped to the pixel range of 0-255, and imaged according to pixels to obtain the logging image as a high-resolution image for storage; the high-resolution image and the corresponding low-resolution image form a paired image dataset; the paired image dataset is augmented to obtain the augmented image dataset;
[0050] In step 1, the stored resistivity data within a 1m depth contains 132 data points, divided into 224 sectors horizontally. To convert the stored data into an image display, the conductivity curve range is first read and mapped to a pixel range of 0-255. Finally, images are formed pixel by pixel, resulting in a high-resolution dataset of 224×132. Due to limitations in telemetry technology, not all stored data can be transmitted to the ground. Therefore, the stored resistivity data is sampled at equal intervals, and the sampled data is imaged to obtain a low-resolution dataset of 56×34. A total of 500 high-resolution-low-resolution image pairs are constructed as the dataset. To expand the dataset, it is horizontally flipped and rotated to obtain 1000 image pairs.
[0051] Step 2. Based on the expanded image dataset obtained in Step 1, build a generative adversarial network model;
[0052] In this embodiment, the specific methods include the following:
[0053] Network structure such as Figure 2 As shown, the generative adversarial network model includes a generator and a discriminator. The generator takes the low-resolution image obtained in step 1 as input and includes convolutional layers, an SE-RRDB module, and an upsampling module. The LR image is the low-resolution image, the SR image is the super-resolution image, and the HR image is the real image. The generator extracts the basic features of the image through the convolutional layers; the SE-RRDB module extracts and fuses the basic features of the image; the upsampling module adjusts the number of channels in the convolutional layers and samples the fused image features to the target output size; the discriminator simultaneously inputs the generated super-resolution image and the original high-resolution image into the discriminator to determine the probability that the image is the real image. Through continuous adversarial interaction between the generator and the discriminator, the final output super-resolution image becomes increasingly closer to the real image.
[0054] like Figure 3 As shown, the low-resolution input image first passes through a convolutional layer to extract primary features. After feature extraction, it passes through multiple stacked SE-RRDB layers to extract image depth features. Then, it is upsampled back to its original size by an upsampling module, and finally outputs a high-resolution image through a convolutional layer.
[0055] like Figure 4 As shown, the RDB submodule includes multiple sequentially connected convolutional layers, and the output of each convolutional layer is concatenated with the output of the previous convolutional layer.
[0056] like Figure 5As shown, the SENet submodule is a module that improves the performance of convolutional neural networks by introducing a channel attention mechanism. The SENet module mainly consists of three parts: compression, activation, and scaling. The compression operation compresses the features of each channel into a single value through global average pooling, thereby achieving global aggregation of feature information. The activation operation recalibrates the features of each channel through a two-layer fully connected network. First, there is a dimensionality reduction operation, followed by a dimensionality increase operation. The scaling operation applies the weights generated by the activation operation to each channel of the input feature map, achieving channel-level feature recalibration.
[0057] like Figure 6 As shown, the discriminator in step 2 includes multiple convolutional layers, with skip connections added between each convolutional layer to preserve low-level features and help the discriminator better capture multi-scale information; used in step 2 to obtain discrimination results from the input high-resolution image and super-resolution image, thereby improving the final effect of super-resolution reconstruction.
[0058] The loss function of the generative adversarial network model is:
[0059] L SR =c1L1+c p L p
[0060] L1 = ||I out -I gt ||1
[0061]
[0062] Among them, L SR Let L1 be the loss function, and c be the pixel loss function. adv To counteract the weighting, L p For the perceptual loss function, c p For perceived weights. I out It is an image generated by the generator, I gt These are true high-resolution images. |||1 represents the L1 norm. N is the number of feature maps, φ i () is the feature map of the i-th layer of the pre-trained VGG network.
[0063] Step 3. Divide the constructed paired image dataset into a training set and a test set in a 9:1 ratio. Use the training set to train the generative adversarial network (GAN) model, and import the test set image dataset into the trained network for testing, performing real-time image super-resolution. Based on the super-resolution results, adjust the hyperparameters of the neural network, including the learning rate, batch size, and loss function weights. Statistically analyze the peak signal-to-noise ratio (PSNR) and structural similarity of the images, compare the image reconstruction performance of the GAN model under different hyperparameters, and find the optimal hyperparameter settings. The network model corresponding to these optimal hyperparameters is the GAN model with the best super-resolution performance, and this optimal setting is saved.
[0064] Step 4. Obtain real-time logging data of resistivity in the target area while drilling, map the real-time logging data of resistivity in the target area to a pixel range of 0-255, image according to pixels, and input it into the generative adversarial network model trained in Step 3 to obtain the real-time image after super-resolution. Figure 7 These are the super-resolution reconstructed images. The first column is the depth channel, recording the measured well depth during the logging process; the second column is the input logging image to be reconstructed; the third column is the reconstructed logging image; and the fourth column is the high-resolution logging image.
[0065] Table 1 presents the results of quantitative evaluation of the reconstructed images using peak signal-to-noise ratio (PSNR) and structural similarity, comparing two super-resolution methods based on generative adversarial networks (GAN) models and one interpolation method with the method of this embodiment (SE-ESRGAN). PSNR represents the degree of similarity between the super-resolution image and the high-resolution image at the pixel level; structural similarity represents the degree of similarity between the super-resolution image and the high-resolution image at the brightness, contrast, and structural levels. As shown in Table 1, both PSNR and structural similarity are optimal, indicating that this method can reconstruct high-quality real-time well logging images.
[0066] Table 1
[0067]
Claims
1. A method for real-time super-resolution reconstruction of resistivity logging images based on generative adversarial networks, characterized in that, Specifically, the following steps are included: Step 1: By reading the logging resistivity data from different regions, the logging resistivity data is sampled at equal intervals and then imaged to obtain the corresponding low-resolution image; the logging resistivity data is mapped to a pixel range of 0-255, and imaged according to pixels to obtain logging images as high-resolution images for storage; the high-resolution image and the corresponding low-resolution image form a paired image dataset; the paired image dataset is augmented to obtain an augmented image dataset; Step 2: Based on the expanded image dataset obtained in Step 1, build a generative adversarial network model; Step 3: Use the expanded image dataset obtained in Step 1 to train and optimize the generative adversarial network model in Step 2, and save the optimal generative adversarial network model. Step 4: Obtain real-time logging data of resistivity in the target area, map the real-time logging data of resistivity in the target area to the pixel range of 0-255, image according to the pixels, and input it into the generative adversarial network model trained in Step 3 to obtain the super-resolution real-time image; compare it with the high-resolution image stored in the logging resistivity data to measure the super-resolution effect. The specific method for building the generative adversarial network model in step 2 is as follows: The generative adversarial network model consists of a generator and a discriminator. The generator takes the low-resolution image obtained in step 1 as input and includes convolutional layers, an SE-RRDB module, and an upsampling module. The generator extracts the basic features of the image through the convolutional layers; the SE-RRDB module extracts and fuses the basic features of the image; the upsampling module adjusts the number of channels in the convolutional layers and samples the fused image features to the target output size; and the low-resolution image is transformed into a super-resolution image output. The discriminator uses a U with spectral normalization. The Net structure evaluates the input high-resolution and super-resolution images through multiple convolutions and skip connections, determines whether the high-resolution and super-resolution images are similar, and feeds the discrimination results back to the generator to guide the generator to generate more realistic images. Set the loss function for the generative adversarial network model to adjust the network parameters of the generator and discriminator until the final generative adversarial network model is obtained. The loss function is: in, For loss function, For pixel loss function, To counter the weight, For the perceptual loss function, For perceived weights; It is an image generated by the generator. It is a true high-resolution image; L1 norm is represented; N is the number of feature maps. It is the pre-trained VGG network. i Feature map of the layer; The process of step 3 is as follows: The paired image dataset from step 1 is divided into a training set and a test set. The generative adversarial network (GAN) model is trained using the training set, and the test set data is imported into the trained GAN model for testing. Super-resolution reconstruction is performed on the low-resolution images. Based on the super-resolution reconstruction results, the hyperparameters of the GAN model are adjusted. The peak signal-to-noise ratio (PSNR) and structural similarity of the images are statistically analyzed to find the optimal hyperparameter settings. The network model corresponding to the optimal hyperparameters is the optimal GAN model, and the optimal GAN model is saved.
2. The method for real-time super-resolution reconstruction of logging-while-drilling resistivity images based on generative adversarial networks according to claim 1, characterized in that, The hyperparameters include learning rate, batch size, and loss function weights.
3. The method for real-time super-resolution reconstruction of logging-while-drilling resistivity images based on generative adversarial networks according to claim 1, characterized in that, The method for finding the optimal hyperparameter settings is as follows: use different combinations of learning rate and batch size to verify the super-resolution reconstruction effect of the generative adversarial network under different training parameters, and select the learning rate and batch size corresponding to the minimum average peak signal-to-noise ratio and structural similarity as the optimal parameters of the generative adversarial network model.
4. A real-time super-resolution reconstruction system for logging resistivity while drilling based on generative adversarial networks, characterized in that, include: The first acquisition module is used to acquire the drilling resistivity storage data from different regions, map the drilling resistivity storage data to a pixel range of 0-255, and image it into a high-resolution image according to the pixels. The drilling resistivity storage data is sampled at equal intervals and then imaged to obtain the corresponding low-resolution image; it is used in step 1 to realize the construction of the dataset. The Generative Adversarial Network (GAN) model building module is used to build a GAN model, which includes a generator and a discriminator; that is, it is used in step 2 to build the GAN model. The generator includes convolutional layers, an SE-RRDB module, and an upsampling module. The convolutional layers extract the basic features of the image; the SE-RRDB module extracts and fuses the basic features of the image. The upsampling module adjusts the number of channels in the convolutional layers and samples the fused image features to the target output size; this is used in step 2 to transform the low-resolution image into a super-resolution image. The SE-RRDB module comprises three sequentially connected RDB sub-modules and one SENet sub-module. The SENet sub-module includes a sequentially connected global average pooling layer, a fully connected layer, and an activation function, which improves the performance of the convolutional neural network by introducing a channel attention mechanism. The RDB sub-module includes multiple sequentially connected convolutional layers, and the output of each convolutional layer is concatenated with the output of the previous convolutional layer. The discriminator, consisting of multiple convolutional layers, incorporates skip connections between each convolutional layer to preserve low-level features, helping the discriminator better capture multi-scale information. In step 2, it is used to obtain discrimination results from the input high-resolution image and super-resolution image, thereby improving the final effect of super-resolution reconstruction. The feedback module sets the loss function of the generative adversarial network model to adjust the network parameters of the generator and discriminator until the final generative adversarial network model is obtained. The network optimization module is used in step 3 to divide the paired image dataset constructed in step 1 into a training set and a test set. The training set is used to train the generative adversarial network (GAN) model, and the test set data is imported into the trained GAN model for testing. Super-resolution reconstruction is performed on low-resolution images. Based on the super-resolution results, the hyperparameters of the GAN model are adjusted. The peak signal-to-noise ratio (PSNR) and structural similarity of the images are statistically analyzed to find the optimal hyperparameter settings. The GAN model corresponding to the optimal hyperparameters is the optimal super-resolution model, and the optimal super-resolution model is saved. The second acquisition module is used in step 4 to acquire the logging-while-drilling resistivity data of the target area and scale it as a logging-while-drilling resistivity image, thereby realizing the acquisition of the real-time logging-while-drilling resistivity image of the target area and the stored logging-while-drilling resistivity image. The super-resolution module is used to input the real-time logging data of the target resistivity while drilling into the super-resolution model to obtain a super-resolution image, which is used in step 4 to realize the super-resolution reconstruction of the real-time logging image of low-resolution resistivity while drilling. The evaluation module is used in step 4 to compare the real-time image after super-resolution reconstruction with the processed and stored data to measure the super-resolution effect.
5. A real-time super-resolution image device for logging resistivity while drilling based on generative adversarial networks, used to implement the method described in any one of claims 1 to 3, characterized in that, include: The system includes a memory and a processor. The memory stores a computer program that, when executed, performs super-resolution reconstruction of real-time images based on the generative adversarial network-based real-time image super-resolution reconstruction method for logging resistivity while drilling, as described in steps 1 to 4, thereby improving image detail information and enhancing image resolution and clarity.
6. A computer-readable storage medium storing a computer program, characterized in that, The computer-readable storage medium provides the function of storing and executing image processing programs, which, when executed by a processor, can achieve efficient and fast real-time super-resolution reconstruction of logging-while-drilling resistivity images based on the generative adversarial network-based real-time image super-resolution reconstruction method for logging-while-drilling resistivity, according to any one of claims 1 to 3.