Geological core image crack and pore identification method and device, electronic equipment and storage medium

Abstract

The invention provides a geological core image crack and pore identification method and device, electronic equipment and a storage medium, and the method comprises the steps: collecting core image data in the same drilling well, carrying out the initialization processing, and inputting the processed images into a model A and a model B; after the model A receives the image, performing preliminary identification on core cracks in the image, outputting a mask for each core crack, segmenting each core crack in the image by using the model B, and enhancing the mask generated by the model A; and adjusting the size of the segmented image to be adaptive to the format and size of the model A, inputting the segmented image into the model A and the model B again for iterative training until the core fracture recognition result in the trained image tends to be stable, and outputting core fracture recognition result data after training. According to the method, the model A and the model B are combined, rapid and accurate identification of the crack holes in the rock core image is achieved, and time and energy in the manual segmentation and calculation process are saved.

Description

Technical Field

[0001] This invention belongs to the field of geological core image processing technology, and particularly relates to a method, device, electronic device and storage medium for identifying cracks and pores in geological core images. Background Technology

[0002] Core images are visual records of cylindrical rock samples drilled from underground rock formations. They provide detailed information on the surface and structural features of the core, enabling detailed observation and analysis of its microstructure and characteristics. This information can also serve as a substitute for physical cores. Furthermore, with the rapid development of computer vision and machine learning, core images are playing a crucial role in oil and gas field exploration and development. These images provide high-resolution three-dimensional representations of the core and its fracture characteristics, including detailed information on fracture pore size, roughness, orientation, and spacing. The distribution patterns of rock fractures can reveal the characteristics of reservoir oil and gas flow. In this process, the results of fracture segmentation and fracture pore size calculations directly affect the accuracy and precision of the exploration of reservoir oil and gas flow characteristics.

[0003] In the current technology, fracture segmentation and fracture aperture calculation are still in the stage of manual segmentation and measurement calculation. This is time-consuming and costly. If the manual measurement and calculation are incorrect, the data on the oil and gas flow characteristics of the exploration reservoir will have significant differences in accuracy and precision. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by providing a method, apparatus, electronic device, and storage medium for identifying fractures and pores in geological core images. It utilizes two image analysis techniques to segment fractures based on geological core images, achieving accurate fracture identification and automatic calculation of fracture pore size. This reduces the time and effort required for manual segmentation and calculation, improves the accuracy and precision of measurement data, and thus promotes research on fracture characterization processes. This invention is achieved through the following technical solution:

[0005] In one aspect of this invention, a method for identifying fractures and pores in geological core images is provided, the method comprising:

[0006] Step 1: Collect core image data from the same well and perform initialization processing. Input the processed images into Model A and Model B.

[0007] Furthermore, the collection and initialization of core images from the same wellbore specifically includes:

[0008] Step 101: Enhance the image quality of low-resolution images in the core images and adjust all core images to a format and size suitable for Model A;

[0009] Step 102: Perform artificial intelligence recognition and verification on the adjusted images to ensure that the crack depiction in each image is close to a uniform result;

[0010] Step 103: Input the verified image into Model A.

[0011] Step 2: After receiving the image, Model A performs preliminary identification of the core cracks in the image and outputs a mask for each core crack in the image. Model B is then used to segment each core crack in the image to enhance the mask generated by Model A.

[0012] Step 3: Use the Microsoft COCO image dataset to adjust the size of the segmented images to fit the format and size of Model A, and then input them into Model A and Model B again for iterative training until the core crack recognition results in the training images tend to be stable. Output the trained core crack recognition result data.

[0013] Furthermore, the core fracture pore size is calculated based on the output core fracture identification results after training, and qualitative and quantitative evaluation is performed by combining the actual core fracture pore size on the ground. This facilitates the optimization of the core fracture identification effect and provides data support for subsequent geological analysis.

[0014] The step of calculating the core fracture pore size based on the output trained core fracture identification results data, and combining it with the actual core fracture pore size on the ground for qualitative and quantitative evaluation, in order to optimize the core fracture identification effect, also includes:

[0015] A visual inspection has confirmed that the cracks were correctly detected.

[0016] All detected core cracks were compared and analyzed with actual core cracks on the ground, reasonable results were retained, and outliers with large differences were removed.

[0017] By using reasonable results data, the error between the detected core fracture pore size and the actual core fracture pore size on the ground is calculated to achieve quantitative analysis.

[0018] Based on the analysis results, the measurement model was improved to optimize the identification effect of core fractures and provide data support for subsequent geological analysis.

[0019] In another aspect of this invention, a device for identifying fractures and pores in geological core images is provided. The device includes a data collection and preprocessing module, a deep learning model processing module, an iterative training module, and a result output and evaluation module, wherein:

[0020] The data collection and preprocessing module is responsible for collecting core image data from the same well and performing initialization processing on this image data.

[0021] The deep learning module processing module is used to perform preliminary recognition on the preprocessed image using Model A, identify the rock core cracks in the image, and output a corresponding mask for each crack. Then, Model B is used to perform fine segmentation on each crack mask output by Model A to enhance the crack recognition effect.

[0022] The iterative training module is used to use the Microsoft COCO image dataset as a reference to adjust the segmented images to fit the format and size of Model A. The adjusted images are then input into Model A and Model B for iterative training. This process is repeated until the core crack identification results in the training images tend to stabilize.

[0023] The result output and evaluation module is used to acquire and output the core fracture identification result data after training, calculate the pore size of the core fracture based on this data, combine it with the actual core fracture pore size on the ground, perform qualitative and quantitative evaluation of the identification results, optimize the core fracture identification effect based on the evaluation results, and provide data support for subsequent geological analysis.

[0024] In another aspect of the present invention, an electronic device is provided, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of a geological core image fracture and pore identification method as described above.

[0025] In another aspect of the present invention, a computer-readable storage medium is provided, wherein a computer program is stored on the storage medium, and when the computer program is executed by the processor, it is used to implement the steps of a geological core image fracture and pore identification method as described above.

[0026] Compared with the prior art, the present invention has the following advantages:

[0027] The present invention provides a method for identifying fractures and pores in geological core images. By integrating advanced data processing and deep learning technologies, this method achieves efficient and accurate identification of fractures and pores in core images. This method not only improves the accuracy and efficiency of identification but also provides strong data support for geological exploration and analysis, contributing to the further development of geological exploration technology.

[0028] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart of a method for identifying cracks and pores in geological core images is shown.

[0031] Figure 2 A structural diagram of a geological core image fracture and pore identification device is shown.

[0032] Figure 3 An experimental architecture diagram of a method for identifying cracks and pores in geological core images is shown. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] To better understand the implementation scheme of the present invention, in the present invention, model A is preferably the Mask-RCNN model, and model B is preferably the PointRend model.

[0035] In one embodiment, please refer to Figure 1 , Figure 1 A flowchart of a method for identifying fractures and pores in geological core images is shown. The method includes the following steps:

[0036] Step S1: Collect core image data from the same well and perform initialization processing. Input the processed images into the Mask-R CNN model.

[0037] Furthermore, the process of collecting and initializing core images from the same wellbore specifically includes the following steps:

[0038] Step S101: Enhance the image quality of low-resolution images in the core images and adjust all core images to a format and size suitable for the Mask-R CNN model;

[0039] Furthermore, the image quality enhancement for low-resolution images includes blurring, sharpening, changing brightness, changing contrast, and horizontal / vertical flipping.

[0040] Step S102: Perform artificial intelligence recognition and verification on the adjusted images to ensure that the crack depiction in each image is close to a uniform result;

[0041] Furthermore, the AI ​​recognition and verification of the adjusted image should be performed more than or equal to 1200 times.

[0042] Step S103: Input the verified image into the Mask-R CNN model.

[0043] Step S2: After receiving the image, the Mask-R CNN model performs preliminary identification of the core cracks in the image and outputs a mask for each core crack in the image. The PointRend model is used to segment each core crack in the image to enhance the mask generated by the Mask-R CNN model.

[0044] Furthermore, the segmentation of each core fracture in the image to enhance the mask generated by the Mask-RCNN model involves: based on the initial identification of core fractures in the image using the Mask-RCNN model, the core fractures in the image are cropped into three image regions from largest to smallest to enhance the mask for further detailed characterization; during the cropping process, it is necessary to ensure that each image region after cropping has a 5% overlap ratio of core fracture height.

[0045] Step S3: Using the Microsoft COCO image dataset, adjust the size of the segmented images to fit the format and size of the Mask-R CNN model, and input them again into the Mask-R CNN model and PointRend model for iterative training until the core crack recognition results in the training images tend to be stable, and output the trained core crack recognition result data.

[0046] Furthermore, the size of the segmented images is adjusted using the Microsoft COCO image dataset to suit the format and size of the Mask-R CNN model, specifically to a size greater than or equal to 1333 pixels.

[0047] Furthermore, the number of iterative training iterations should be greater than or equal to 7000; the core crack identification results tend to be stable, specifically meaning that the image core crack identification results are uniform and the detail error does not exceed 5%.

[0048] Furthermore, based on the output training results of core fracture identification, the core fracture pore size is calculated, and qualitative and quantitative evaluations are performed by combining this data with the actual core fracture pore size on the ground. This facilitates the optimization of core fracture identification performance and provides data support for subsequent geological analysis. The process includes the following steps:

[0049] First, a visual inspection was conducted to confirm the correct cracks that were detected.

[0050] Next, all detected core cracks were compared and analyzed with actual core cracks on the ground, retaining reasonable results and removing outliers with large discrepancies.

[0051] Then, using reasonable result data, the error between the detected core fracture pore size and the actual core fracture pore size on the ground is calculated to achieve quantitative analysis;

[0052] Finally, based on the analysis results, the measurement model was improved to optimize the identification effect of core fractures and provide data support for subsequent geological analysis.

[0053] In one embodiment, please refer to Figure 2 , Figure 2 A structural diagram of a geological core image fracture and porosity identification device is shown. The device includes a data collection and preprocessing module, a deep learning model processing module, an iterative training module, and a result output and evaluation module, wherein:

[0054] The data collection and preprocessing module is responsible for collecting core image data from the same well and performing initialization processing on this image data.

[0055] The deep learning module processing module is used to perform preliminary recognition on the preprocessed image using the Mask-R CNN model, identify the rock core cracks in the image, and output a corresponding mask for each crack. The PointRend model is used to perform fine segmentation on each crack mask output by the Mask-R CNN model to enhance the crack recognition effect.

[0056] The iterative training module is used to use the Microsoft COCO image dataset as a reference to adjust the segmented images to fit the format and size of the Mask-R CNN model, and then input the adjusted images back into the Mask-R CNN model and the PointRend model for iterative training. This process is repeated until the core crack identification results in the training images tend to be stable.

[0057] The result output and evaluation module is used to acquire and output the core fracture identification result data after training, calculate the pore size of the core fracture based on this data, combine it with the actual core fracture pore size on the ground, perform qualitative and quantitative evaluation of the identification results, optimize the core fracture identification effect based on the evaluation results, and provide data support for subsequent geological analysis.

[0058] In one embodiment, an electronic device is provided, including a memory and a processor. The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of a geological core image fracture and pore identification method as described above.

[0059] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when executed by the processor, is used to implement the steps of a geological core image fracture and porosity identification method as described above.

[0060] Experimental example:

[0061] To better understand this invention, experimental examples are provided for further explanation. Please refer to [link / reference]. Figure 3 , Figure 3 An experimental architecture diagram of a method for identifying fractures and pores in geological core images is shown below:

[0062] Core images from two boreholes in an oilfield were selected, initially totaling 80 images, including 160 natural fractures. Image enhancement was used to expand the images to 1000, with 800 images used for training and 200 for validation. The fracture images labeled for training, validation, and testing were resized to a height of 811–2905 pixels and a width of 356–8980 pixels, resulting in a resolution of 2.6–9.0 pixels / mm. To build the second-stage segmentation model, three smaller images were created from each fracture in the training and validation stages, resized to a height of 144–1057 pixels and a width of 74–3457 pixels. Before inputting the images into the model at each segmentation stage, they were resized to the default size for Mask R-CNN. After resizing, the images had different resolutions depending on the original resolution and the image size controlling the resizing ratio.

[0063] The Mask R-CNN model receives the Mask R-CNN image and outputs a mask for each crack, similar to a typical instance segmentation model. Then, the PointRend model is used to segment small regions of the image, enhancing the mask generated by the Mask R-CNN model. Based on the crack locations obtained from the Mask R-CNN model, each crack in the image is divided into three regions, thus achieving enhancement. The cropped image is then re-segmented using the Mask R-CNN model, and the generated high-resolution mask is used to replace the original mask of the Mask R-CNN model, undergoing ≥7000 iterations.

[0064] Visual inspection of the generated results was performed to confirm correctly detected cracks, and outliers were removed from subsequent analysis by comparing all detected cracks with actual ground cracks. Visual inspection evaluated the quality of segmentation, including comparing single-level and two-level segmentation, and comparing results from standard Mask R-CNN and Mask R-CNN+PointRend. Quantitative analysis was achieved by calculating the relationship between actual and predicted cracks and measuring the impact of model improvements on the accuracy of crack aperture calculation. Compared to manual segmentation, the above method improved accuracy by 55%. With the enhanced crack identification process, the aperture error distribution was significantly improved.

[0065] Overall, comparing the standard Mask R-CNN with Mask R-CNN+PointRend reveals that the latter provides more accurate and detailed segmentation, but is affected by overlap and adjacent gaps.

[0066] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying fractures and pores in geological core images, characterized in that: Collect core image data from the same well and perform initialization processing. Input the processed images into model A and model B. After receiving the image, Model A performs preliminary identification of the core fractures in the image and outputs a mask for each core fracture in the image. Model B is then used to segment each core fracture in the image to enhance the mask generated by Model A. The size of the segmented image is adjusted to fit the format and size of Model A, and then input into Model A and Model B again for iterative training until the core crack recognition results in the training image tend to be stable. The trained core crack recognition result data is then output.

2. The method for identifying fractures and pores in geological core images as described in claim 1, characterized in that, The process of collecting and initializing core images from the same wellbore also includes: Image quality enhancement was performed on low-resolution images in the core images, and all core images were adjusted to a format and size suitable for Model A. The adjusted images are then subjected to artificial intelligence recognition and verification to ensure that the crack depiction in each image is close to a uniform result. Input the verified image into model A.

3. The method for identifying fractures and pores in geological core images as described in claim 2, characterized in that: The low-resolution images in the core images are enhanced in terms of image quality, including blurring, sharpening, changing brightness, changing contrast, and horizontal and vertical flipping. The adjusted image is then subjected to artificial intelligence recognition and verification.

4. The method for identifying fractures and pores in geological core images as described in claim 1, characterized in that: The segmentation of each core fracture in the image is specifically based on the location of the core fracture obtained by preliminary identification of the core fracture in the image using the A model, and then the image is cropped into three image regions from largest to smallest.

5. The method for identifying fractures and pores in geological core images as described in claim 1, characterized in that: The core fracture identification results tend to be stable, specifically meaning that the image core fracture identification results are uniform and the detail error is less than or equal to 5%.

6. A method for identifying fractures and pores in geological core images as described in any one of claims 1 to 5, characterized in that, Also includes: The core fracture diameter is calculated based on the core fracture identification results after training, and qualitative and quantitative evaluation is performed by combining the actual core fracture diameter on the ground.

7. The method for identifying fractures and pores in geological core images as described in claim 6, characterized in that, The step of calculating the core fracture pore size based on the output core fracture identification results data after training, and performing qualitative and quantitative evaluation by combining it with the actual core fracture pore size on the ground, also includes: Conduct a visual inspection to confirm the correct cracks detected; All detected core fractures were compared and analyzed with actual core fractures on the ground, and reasonable results were retained. By using reasonable results data, the error between the detected core fracture pore size and the actual core fracture pore size on the ground is calculated to achieve quantitative analysis. Based on the analysis results, the measurement model was improved to optimize the identification effect of core fractures.

8. A device for identifying fractures and pores in geological core images, characterized in that: The device includes a data collection and preprocessing module, a deep learning model processing module, an iterative training module, and a result output and evaluation module, wherein: The data collection and preprocessing module is responsible for collecting core image data from the same well and performing initialization processing on this image data. The deep learning module processing module is used to perform preliminary recognition on the preprocessed image using model A, identify the rock core cracks in the image, and output a corresponding mask for each crack. Then, model B is used to perform fine segmentation on each crack mask output by model A to enhance the crack recognition effect. The iterative training module is used to adjust the segmented image to fit the format and size of model A, and then input the adjusted image back into model A and model B for iterative training. This process is repeated until the core crack identification results in the training image tend to be stable. The result output and evaluation module is used to acquire and output the core fracture identification result data after training, calculate the pore size of the core fracture based on the data, combine it with the actual core fracture pore size on the ground, perform qualitative and quantitative evaluation of the identification results, and optimize the core fracture identification effect based on the evaluation results.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, the processor performs a method for identifying fractures and pores in geological core images as described in any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements a method for identifying cracks and pores in geological core images as described in any one of claims 1 to 6.

Images