Electrical imaging sinusoidal fracture identification method and device based on hybrid curve fitting algorithm, equipment and medium

Abstract

The application discloses a mixed curve fitting algorithm-based electric imaging sinusoidal fracture identification method and device, equipment and medium, which comprises the following steps: collecting and cleaning structured data; converting the cleaned structured data into an electric imaging logging image; inputting the electric imaging logging image into an image segmentation model to output a binary image; performing morphological operation on the binary image; finding the boundaries of all continuous non-0 pixel regions in the profile of the binary image after the morphological operation to obtain a profile set; performing multi-stage screening on the profile set to remove noise and invalid profiles; performing mixed curve fitting on the screened profile set by using a sinusoidal function model in combination with periodic continuation rules and amplitude constraint physical constraints to generate a complete-period sinusoidal fracture curve; and superimposing the sinusoidal fracture curve and the electric imaging logging image to complete fracture identification. Therefore, the application can be widely applied to logging evaluation of fractured reservoirs such as carbonate rocks, shales and tight sandstones.

Description

Technical Field

[0001] This invention relates to the fields of oil well logging, geological engineering and image processing technology, and particularly to the field of well logging interpretation technology in oil and gas exploration and development. Specifically, it relates to a method, device, equipment and medium for identifying sinusoidal fractures using electrical imaging based on a hybrid curve fitting algorithm. Background Technology

[0002] In oil and gas exploration and development, the degree of reservoir fracture development directly affects the storage space and seepage channels for oil and gas. Electrical imaging logging, due to its high resolution and high wellbore coverage, can clearly reflect geological features such as wellbore lithology, structural fractures, and pores, and has become a core technology for identifying and evaluating reservoir fractures. In electrical imaging logging images, since the wellbore is cylindrical, the image is a two-dimensional planar diagram unfolded using polar coordinates. Therefore, when high-angle or inclined fractures cut through the wellbore, they will exhibit a typical sinusoidal waveform (i.e., sinusoidal fractures) on the two-dimensional unfolded diagram. Accurately extracting the parameters of these sinusoidal curves (such as extreme point coordinates, amplitude, and phase) is the prerequisite and foundation for calculating fracture attitude (dip angle, strike) and for subsequent evaluation of fracture density and aperture.

[0003] Currently, existing technologies for identifying and extracting sinusoidal fractures in electro-optical imaging images mainly employ the following methods: 1) Manual interactive picking method: Interpreters manually trace the fracture trajectory on the image. Although this method has high accuracy, it is extremely inefficient, highly subjective, and cannot meet the needs of rapid processing of massive amounts of logging data. 2) Methods based on traditional image processing and Hough transform: Fracture pixels are first extracted through edge detection (such as the Canny operator), and then Hough transform is used for voting detection in the parameter space. However, electro-optical imaging images have complex backgrounds, and edge detection easily generates a large number of discontinuous pseudo-edges and isolated noise points. Traditional Hough transform is not only computationally intensive and time-consuming when processing such discontinuous, high-noise sinusoidal curves, but it is also prone to missed detections and false detections. 3) Methods based on single traditional curve fitting: After initially extracting contour points, a single algorithm such as the standard least squares method is directly used for sinusoidal curve fitting. Due to the presence of conductive patches or bedding planes in the logging environment, a large number of outliers are inevitably mixed into the initially extracted contour point set. The single least squares method is extremely sensitive to outliers. The fitting process is easily "shortened" by outliers, resulting in severe distortion of the fitted sine curve, which cannot truly reflect the crack morphology and thus causes errors in the attitude calculation.

[0004] In summary, existing automatic crack identification methods based on electro-imaging struggle to achieve a good balance between noise resistance, robustness, and fitting accuracy. When faced with poor image quality, discontinuous crack features, or complex geological interference, traditional methods fail to balance computational efficiency and fitting accuracy. Therefore, there is an urgent need for an identification method that can effectively eliminate abnormal interference, exhibit strong robustness to discontinuous contours, and accurately fit the true sinusoidal crack trajectory. Summary of the Invention

[0005] The present invention aims to at least solve one of the technical problems existing in the prior art. Therefore, in response to the above-mentioned problems, the object of the present invention is to provide a method, apparatus, device, and medium for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm, overcoming the shortcomings of existing methods in terms of noise resistance, discontinuous fitting ability, and computational efficiency.

[0006] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm, comprising: Structured data acquisition and cleaning; The cleaned structured data is converted into electrical imaging logging images, which include high-conductivity fractures, high-resistance fractures, and / or feather-shaped induced fractures. The electrical imaging logging image is input into a pre-trained image segmentation model, which outputs a binary image. In the binary image, the pixel value of the fracture region is 255 and the background is 0. Perform morphological operations on binary images; Find the boundaries of all consecutive non-zero pixel regions in the binary image contour after morphological operations to obtain the contour set; Perform multi-level filtering on the contour set to remove noise and invalid contours; A sinusoidal function model is used, combined with the physical constraints of periodic extension rules and amplitude constraints, to perform mixed curve fitting on the screened contour set and generate a sinusoidal crack curve with a complete cycle. The sinusoidal fracture curve and the electrical imaging logging image are superimposed to complete the fracture identification.

[0007] Some possible implementations of structured data acquisition and cleaning include: Data was simultaneously acquired using a downhole high-resolution multi-plate array micro-resistivity imager, including resistivity matrix, natural gamma curve, well inclination and azimuth, mud resistivity, and well temperature data. The collected data is cleaned and corrected; The cleaned and corrected data are stored in depth order as a format file as a standardized base dataset.

[0008] In some possible implementations, the cleaned structured data is converted into electrical imaging logging images, including: Convert the resistivity matrix into an electrical imaging logging image; Based on the well inclination azimuth information, the electrical imaging logging images are azimuth corrected so that the top of the electrical imaging logging images points to geographic north or magnetic north. Among them, the fracture characteristics in the electrical imaging logging images are as follows: high-conductivity fractures appear as dark or black linear / curved features; high-resistance fractures appear as light or white linear features; and feather-shaped induced fractures appear as complex curved shapes resembling tree branches or feathers.

[0009] In some possible implementations, the image segmentation model is trained using an improved YOLOv11seg deep learning segmentation model. This improved YOLOv11seg model integrates a C3K2 enhanced feature extraction module, a SPFF multi-scale fusion module, an adaptive anchor box mechanism, and an EIOU loss function. The training process of the image segmentation model is as follows: Several electro-imaging image samples of labeled crack regions were collected as a training set. The labeled crack regions included high-conductivity cracks, high-resistance cracks, and feather-induced cracks. The weights pre-trained on the COCO dataset were used as initial parameters for fine-tuning through transfer learning. The optimizer used was AdamW with an initial learning rate of 0.001, momentum of 0.9, weight decay of 0.0005, batch size of 8, and training for 50 epochs. After training, the image segmentation model was obtained.

[0010] In some possible implementations, morphological operations are performed on the binary image, including: opening and closing operations are performed on the binary image. The opening operation is used to remove isolated foreground noise and cut off small spiky connections; the closing operation is used to fill the tiny holes inside the cracks and connect broken segments caused by discontinuity in segmentation; the execution order and number of iterations are such that the opening operation is performed once first, followed by the closing operation once. In areas with low foreground probability, the number of dilation operations in the closing operation is reduced to avoid spurious connections, resulting in an optimized binary image.

[0011] In some possible implementations, the contour set is subjected to multi-level screening to remove noise and invalid contours. Specifically, the contours of all acquired continuous regions are subjected to dual validity screening: first, threshold screening based on area; second, grouping and merging based on the same horizontal line, thereby integrating multiple contour segments belonging to the same crack into a set of points to be fitted. (a) Area-based threshold filtering: Calculate the area of ​​each contour and retain only valid contours with an area greater than or equal to a set number of pixels; (b) Grouping and merging based on the same horizontal line: Determining if two contours are on the same horizontal line: If the following conditions are met, then it is determined to be "the same horizontal line": Where T is the threshold. , The ordinate of the center of the two contours; Grouping: Let the initial grouping set be... Execution: Check the collection Does the group already exist? and The first outline If a group satisfies the condition of "being on the same horizontal line", then such a group exists. Then Join this group ,Right now If it does not exist, create a new group. This results in the final set of contour groups, with each group corresponding to a candidate crack.

[0012] In some possible implementations, a sinusoidal function model is used, combined with physical constraints such as periodic extension rules and amplitude constraints, to perform hybrid curve fitting on the screened contour set, generating a sinusoidal crack curve with a complete cycle, including: Periodic continuation rule: Forced periodic continuation, fixed period T = width, angular frequency B = If the fracture profile does not fully represent a cycle in the image, the cycle is extended based on the image width. That is, the cycle T is set to equal the image width, and the complete cycle curve is supplemented in the extrapolated part of the image by the periodicity of the sine function, so as to ensure that the fitted curve covers the entire logging image range. Physical constraints of amplitude constraint: (a) Amplitude is limited to A≤ preset value; (b) Period is limited to a complete period within 0~width pixels, and if it is less than a period, it is extended to a period. Crack curve fitting: Based on the periodic characteristics of cracks in electro-imaging images, a sine function Y=A·sin(B·x+C)+D is used for curve fitting, where A is the amplitude, B is the angular frequency, C is the initial phase, and D is the baseline offset.

[0013] Secondly, the present invention also provides an electro-imaging sinusoidal crack identification device based on a hybrid curve fitting algorithm, comprising: The data acquisition unit is configured for structured data acquisition and cleaning; The data processing unit is configured to convert the cleaned structured data into electrical imaging logging images, wherein the electrical imaging logging images include high-conductivity fractures, high-resistance fractures and / or feather-shaped induced fractures; The image segmentation unit is configured to input electrical imaging logging images into a pre-trained image segmentation model and output a binary image, wherein the pixel value of the fracture region in the binary image is 255 and the background is 0; The image binarization unit is configured to perform morphological operations on the binary image; The contour acquisition unit is configured to find the boundaries of all consecutive non-zero pixel regions in the contour of the binary image after morphological operations, and obtain a contour set. The contour filtering unit is configured to perform multi-level filtering on the contour set to remove noise and invalid contours. The curve fitting unit is configured to use a sinusoidal function model, combined with physical constraints of periodic extension rules and amplitude constraints, to perform mixed curve fitting on the screened contour set, generating a sinusoidal crack curve of the complete cycle. The fracture identification unit is configured to overlay sinusoidal fracture curves and electrical imaging logging images to identify fractures.

[0014] Thirdly, the present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to enable the processor to perform the method described thereon.

[0015] Fourthly, the present invention also provides a computer-readable storage medium for storing one or more programs, characterized in that the one or more programs include computer instructions for causing a computer to perform any of the methods described herein.

[0016] Compared to existing crack identification methods based on Hough transform or single least squares fitting, this invention employs a hybrid curve fitting algorithm. Through the synergistic effect of sine function model construction, periodic extension rules, and physical constraint control, it achieves the following beneficial effects: 1. Based on the periodic extension mechanism with a fixed period forced angular frequency, the completeness and robustness of fitting incomplete fracture profiles are significantly improved: This invention addresses the problem that fracture profiles in electro-imaging images often lack part of the period due to wellbore truncation, image boundaries, or local development. It innovatively introduces a periodic extension rule with the image width as the fixed period, so that the fitting process no longer depends on the complete coverage of the profile points, which significantly improves the fitting probability.

[0017] 2. Introducing physical constraints such as amplitude ensures the physical rationality of the fitting results and the adaptability of the image boundaries: This invention applies physical constraints such as amplitude limit A ≤ 200 pixels in the sine curve fitting. This constraint is based on the physical prior of electrical imaging logging: the fracture dip angle is usually no more than 85°, corresponding to the maximum vertical fluctuation of the sine curve being less than half the image height (the typical electrical imaging image height is 800~1200 pixels, and 200 pixels corresponds to a reasonable dip angle range of approximately 15°~30°). The purpose of this constraint is to prevent the fitting algorithm from generating unreasonable curves with excessive amplitude and waveform distortion when there are outliers or noise in the contour point set. Comparative experiments show that, with the addition of 10% random outliers, the relative error between the fitting amplitude of this invention and the true value is 5.2%, while the error of the traditional unconstrained least squares method is as high as 28.7%. At the same time, by checking the boundary between the baseline D and the amplitude A, it is ensured that the fitted curve is always within the effective height of the image, avoiding errors in attitude calculation or abnormal image display caused by the curve going out of bounds.

[0018] 3. Based on the combination of sine function model and robust fitting algorithm, high-precision crack parameter extraction with resistance to outliers is achieved: This invention retains the sine function model as the theoretical basis for crack morphology, which accurately describes the geometric features of inclined cracks in electro-imaging images.

[0019] 4. Full automation from contour extraction to fitting output significantly improves processing efficiency: This invention seamlessly integrates multi-contour extraction, validity screening (area ≥ 100 pixels, grouped along the same horizontal line), and hybrid curve fitting without manual intervention. Compared to the manual picking method in existing technologies, the processing time for a single well in this invention is reduced from 4-6 hours manually to less than 5 minutes, improving efficiency by approximately 50 times. Compared to the Hough transform, the average computation time of this invention is lower than that of the Hough transform (approximately 1.2 seconds per fracture), and its processing performance for high-density fracture segments is more stable. This feature directly meets the industrial demand for rapid processing of massive logging data.

[0020] 5. Final effect display and parameter output provide intuitive and quantifiable fracture interpretation results: This invention displays the fitted complete periodic sine curve in a bright color on the original electrical imaging image, and outputs the fracture type (high-conductivity fracture / high-resistance fracture / feather-induced fracture) and quantitative parameters (amplitude, phase, baseline, which can be converted into dip angle and strike). The purpose of this feature is to enable interpreters to intuitively verify the accuracy of the fitting results and provide directly usable digital parameters for subsequent fracture density, aperture, and reservoir productivity evaluation, eliminating the black box problem of "fitting results but not knowing whether they are reliable" in traditional methods.

[0021] In summary, this invention relates to image processing and pattern recognition technologies such as structured acquisition, target segmentation, morphological processing, and sinusoidal curve fitting of electro-imaging logging images. In particular, it integrates deep learning segmentation and hybrid curve fitting to achieve automatic detection and quantitative characterization of wellbore fractures. The above-mentioned links are interconnected and progressively advanced, forming a fully automatic, robust, and high-precision electro-imaging sinusoidal fracture identification technology solution. This overcomes the shortcomings of existing methods in terms of noise resistance, discontinuous fitting ability, and computational efficiency. It can be widely applied to well logging evaluation of fractured reservoirs such as carbonate rocks, shale, and tight sandstone, providing a reliable technical means for fracture parameter extraction and reservoir productivity prediction. Attached Figure Description

[0022] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings: Figure 1 This is a flowchart of the electro-imaging sinusoidal crack identification method of the present invention; Figure 2 This is the original image of Example 1 of the high-resistance seam test of the present invention; Figure 3 This is a binarized high-resistance seam image after YOLOv11seg segmentation calculation in Embodiment 1 of the present invention; Figure 4 This is a scatter plot of the distribution after the multi-contour extraction rule is applied in Embodiment 1 of the present invention; Figure 5 This is a schematic diagram of the sine curve of the high-resistance slit in Embodiment 1 of the present invention after fitting through constraint control, periodic extension, etc. Figure 6 The diagram shows a comparison of the effects of constraint control and periodic extension on the high-resistance slit in Embodiment 2 of the present invention. The left diagram shows the slit without constraint control and periodic extension, and the right diagram shows the slit with constraint control and periodic extension. Figure 7 This is the final result diagram after plotting the original image and the high-resistance seam recognition and fitting curve of Embodiment 1 of the present invention; Figure 8 The images are the original image of the feather-induced slit embodiment 3 of the present invention and the binarized feather-induced slit image after YOLOv11seg segmentation calculation; Figure 9 The images show the multi-contour extraction results and the final feather-induced slit fitting effect of Example 3 of the present invention. Figure 10 This is a structural diagram of the electronic device of the present invention. Detailed Implementation

[0023] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a specific order as described or illustrated, unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0024] Although terms such as first, second, third, etc., may be used in this document to describe multiple elements, components, regions, layers, and / or segments, these elements, components, regions, layers, and / or segments should not be limited by these terms. These terms may be used only to distinguish one element, component, region, layer, or segment from another. Unless the context clearly indicates otherwise, terms such as "first," "second," and other numerical terms used herein do not imply order or sequence. Therefore, the first element, component, region, layer, or segment discussed below may be referred to as the second element, component, region, layer, or segment without departing from the teachings of the exemplary embodiments.

[0025] For ease of description, spatial relative terms may be used in the text to describe the relationship of one element or feature relative to another element or feature as shown in the figure. These relative terms include, for example, "inside," "outside," "middle," "outer," "below," "above," etc. Such spatial relative terms are intended to include different orientations of the device in use or operation, other than those depicted in the figure.

[0026] To address the following shortcomings in existing sinusoidal crack identification methods using electrical imaging: low efficiency and strong subjectivity of manual picking; poor noise resistance of traditional Hough transform and easy to miss discontinuous cracks; and sensitivity of single least squares fitting to outliers and inability to handle incomplete crack contours. This invention discloses a method, apparatus, device, and medium for identifying sinusoidal fractures using electro-imaging based on a hybrid curve fitting algorithm. The method includes: structured data acquisition and cleaning; converting the cleaned structured data into electro-imaging logging images, wherein the electro-imaging logging images include high-conductivity fractures, high-resistance fractures, and / or feather-shaped induced fractures; inputting the electro-imaging logging images into a pre-trained image segmentation model to output a binary image, wherein the pixel value of the fracture region in the binary image is 255 and the background is 0; performing morphological operations on the binary image; finding the boundaries of all continuous non-zero pixel regions in the contours of the morphologically operated binary image to obtain a contour set; performing multi-level filtering on the contour set to remove noise and invalid contours; using a sinusoidal function model, combined with physical constraints of periodic extension rules and amplitude constraints, to perform hybrid curve fitting on the filtered contour set to generate a sinusoidal fracture curve with a complete period; and superimposing the sinusoidal fracture curve and the electro-imaging logging image to complete fracture identification. Therefore, this invention can automatically and robustly identify various sinusoidal fractures, effectively suppress anomaly interference, and significantly improve the fitting accuracy of incomplete contours.

[0027] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0028] Example 1: The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm provided in this example is as follows: S1. Structured data acquisition and cleaning.

[0029] In this embodiment, a structured data acquisition and cleaning process for the field of acoustic-electric imaging is constructed. Key parameters such as resistivity, natural gamma, and wellbore inclination are simultaneously acquired using a downhole multi-sensor array. The acquired data undergoes cleaning processes such as temperature correction, noise filtering, and depth alignment to output a standardized base dataset, providing high-quality input for subsequent image conversion. The specific process is as follows: S11. Using a downhole high-resolution multi-plate array microresistivity imager (such as EMI, FMI, or similar instruments), the following data are acquired synchronously at fixed depth sampling intervals (typically 0.1 inches or 0.2 inches): Resistivity matrix: The apparent resistivity values ​​measured by multiple plates (e.g., typically 6 to 8 plates, each containing 24 button electrodes, for a total of 144 to 192 electrodes, but not limited to this example) generate a resistivity vector for each depth point.

[0030] Natural gamma curve: Measured using a natural gamma logging tool, used for depth correction and lithological comparison.

[0031] Well inclination and azimuth: Measured using a triaxial accelerometer and magnetometer for image orientation.

[0032] Mud resistivity and well temperature: Measurements were performed using a four-electrode downhole mud conductivity sensor and a platinum resistance temperature sensor for resistivity environmental calibration.

[0033] All the above data are indexed by depth and encapsulated into structured frames in the order of sampling. Each frame contains a timestamp, depth value, resistivity matrix, other auxiliary parameter vectors, and quality flags (such as good plate contact flag and signal saturation flag) to form the raw base data (base data refers to the most original, unprocessed structured measurement data). S12. Perform data cleaning and correction.

[0034] In this embodiment, the following cleaning operations are performed sequentially on the original base data packets: Spike noise removal: For the resistivity depth sequence of each electrode channel, a 3-point sliding median filter is applied, and a threshold (e.g., exceeding the neighborhood mean ± 3 times the standard deviation) is set to identify and remove isolated spikes. For bad channels (more than 5 consecutive invalid data points), the channel is marked as invalid, and linear interpolation of adjacent plates is used to fill in the gaps during subsequent image reconstruction.

[0035] Mud resistivity correction: Based on the measured mud resistivity, the Aziz formula or empirical correction factor is used to eliminate the influence of mud diversion.

[0036] Temperature correction: Based on the readings of the downhole temperature sensor, the measured resistivity is converted to resistivity at a standard temperature (e.g., 25°C) using the temperature-resistivity correction curve calibrated by the instrument at the factory.

[0037] S13. Store the cleaned and calibrated data in depth order as a standardized base dataset: resistivity matrix (depth × number of electrodes), natural gamma vector, confidence weight matrix (obtained by combining instrument status information and quality flags during cleaning and calibration, used to measure the reliability of each measurement data point, its dimension is the same as the resistivity matrix dimension), well inclination azimuth matrix, and metadata (instrument model, well number, sampling interval, etc.).

[0038] S2. Convert structured data into electrical imaging logging images.

[0039] In this embodiment, the cleaned structured data is converted into an electrical imaging logging image. This image retains typical fracture features such as high-conductivity fractures, high-resistance fractures, and feather-shaped induced fractures. The fractures exhibit linear or curved structures with grayscale differences from the background. The specific process is as follows: S21. Convert the resistivity matrix into an electrical imaging logging image.

[0040] In this embodiment, the resistivity matrix is ​​converted into an electrical imaging logging image through resistivity-grayscale mapping: First, logarithmic transformation or percentile normalization is used to map the resistivity values ​​to a grayscale range of 0~255; then, image reconstruction is performed, and the resistivity grayscale values ​​of each depth point are arranged in depth order to form a two-dimensional grayscale image with a width equal to the total number of electrodes (or the fixed number of sectors after interpolation, usually 360 pixels / week) and a height equal to the number of depth sampling points.

[0041] S22. Based on the well inclination azimuth information, perform azimuth correction on the electrical imaging logging image so that the top of the electrical imaging logging image points to geographic north or magnetic north.

[0042] S23. Obtain fracture characteristics in the generated electrical imaging logging images: high-conductivity fractures (fractures filled with conductive fluids) appear as dark or black linear / curved features; high-resistivity fractures (filled with high-resistivity minerals or oil and gas) appear as bright or white linear features; feather-shaped induced fractures appear as complex curved shapes resembling trees or feathers.

[0043] S3. Crack region segmentation based on image segmentation model.

[0044] In this embodiment, the electrical imaging logging image is input into the image segmentation model, which is trained using the YOLOv11seg segmentation model. This model is optimized through transfer learning and can identify three types of fractures in the electrical imaging logging image: high-conductivity fractures (gdf), high-resistance fractures (gzf), and feather-shaped induced fractures (yz). It outputs a binary segmentation result containing the fracture area (fracture area as foreground and non-fracture area as background). The segmentation result is thresholded to obtain a binary image with a fracture area pixel value of 255 and a background value of 0, ensuring that the fracture outline is clearly distinguishable.

[0045] Furthermore, the YOLOv11seg segmentation model is a single-stage end-to-end object detection model that treats the detection task as a regression problem, directly predicting bounding boxes and categories. This embodiment uses an improved YOLOv11seg deep learning segmentation model to perform end-to-end fracture region segmentation on the aforementioned electrical imaging logging images. This model integrates a C3K2 enhanced feature extraction module, a SPFF multi-scale fusion module, an adaptive anchor box mechanism, and an EIOU loss function. It can accurately identify three types of fractures and output binary segmentation results, significantly improving noise resistance and segmentation accuracy in complex geological environments. The backbone network of this model includes the following improved modules: The C3K2 module introduces optional convolution kernel sizes (3×3 or 5×5) on top of the CSP structure, enhancing the ability to extract multi-scale features. For minute fractures with a width of only 1-3 pixels in electrical imaging logging images, a 5×5 convolution kernel is selected to expand the receptive field.

[0046] SPFF module: It adopts spatial pyramid pooling to fuse multi-scale features, optimizes the multi-scale feature fusion process, and effectively improves the detection performance of small targets (such as isolated dissolved holes and microcracks).

[0047] Adaptive anchor box mechanism: Based on the statistical distribution of crack bounding boxes in the training set, the optimal anchor box size is automatically generated by clustering, which reduces the cost of manual parameter tuning; EIOU loss function: This function extends the traditional IoU loss by adding penalty terms based on center point distance and aspect ratio. The formula is as follows:

[0048] in, Indicates the prediction box; These represent the horizontal and vertical coordinates of the center point of the prediction box, as well as the width and height of the prediction box, respectively. Represents a real bounding box; These represent the horizontal and vertical coordinates of the center point of the ground truth bounding box, as well as the width and height of the predicted bounding box; Represents the square of the distance from the center point; This represents the smallest bounding rectangle that completely contains both the predicted bounding box and the ground truth bounding box. This indicates its width and height.

[0049] In this embodiment, the training process of the YOLOv11seg segmentation model is as follows: Several electro-imaging image samples of labeled crack regions (high-conductivity crack gdf, high-resistance crack gzf, and feather-induced crack yz) were collected and divided into training set, validation set and test set in a ratio of 8:1:1.

[0050] Weights pre-trained on the COCO dataset were used as initial parameters for fine-tuning via transfer learning. The optimizer used was AdamW with an initial learning rate of 0.001, momentum of 0.9, weight decay of 0.0005, batch size of 8, and a training duration of 50 epochs. After training, an image segmentation model was obtained. In practice, an electro-imaging image was input into the trained image segmentation model, which outputs a probability map indicating that each pixel belongs to the foreground (crack). Set threshold This converts the probability map into a binary image.

[0051] S4. Morphological processing.

[0052] In this embodiment, morphological operations such as erosion and dilation are performed on the binary image to improve crack connectivity and remove noise. Mathematical morphological processing, including opening and closing operations, is applied to the segmented binary image. Elliptical structuring elements are used to eliminate isolated noise, fill micro-holes, and connect fractured crack segments. The binarized segmented image undergoes erosion and dilation operations to remove residual isolated noise, fill micro-holes within cracks, and connect fractured crack segments caused by discontinuous segmentation. Erosion and dilation are basic mathematical morphological operations, employing a structuring element (kernel) sliding across the binary image, with logical judgments based on the pixel set covered by the structuring element.

[0053] Specifically, the structuring element is chosen to be elliptical, with a core size of k×k. The value of k is set based on the image resolution and crack width: for standard electro-optic images (360 pixels horizontally, 8 pixels per inch vertically), k=3. If the image noise is high or the crack fracture is severe, the value of k can be increased appropriately. Next, opening and closing operations are performed. The opening operation (erosion followed by dilation) is used to remove isolated foreground noise (spots with an area smaller than the structuring element) and to sever fine, burr-like connections. The closing operation (dilation followed by erosion) is used to fill the tiny pores inside the crack and connect broken segments caused by discontinuous segmentation. The execution order and number of iterations are as follows: one opening operation followed by one closing operation. In areas with low foreground probability (such as near bad channels on the electrode), the number of dilation operations in the closing operation can be reduced (e.g., one dilation followed by direct erosion, without unnecessary iterations) to avoid spurious connections, ultimately resulting in an optimized binary image.

[0054] S5, Multi-contour extraction.

[0055] In this embodiment, the core OpenCV function `cv2.findContours()` is used to extract contours from binary images. This function finds the boundaries of all consecutive non-zero pixel regions in a binary image (non-zero pixels are the target, zero pixels are the background). These boundaries are called "contours". A contour is an abstract representation of the shape of a target in an image, consisting of a series of consecutive pixel coordinates, and is the basis for subsequent shape analysis and curve fitting.

[0056] Specifically, the `cv2.findContours()` function from the OpenCV library is used. The optimized binary image is input, the contour retrieval mode is selected to extract only the outermost contour, and the contour approximation method is selected to compress redundant points in the horizontal, vertical, and diagonal directions. Let the input binary image be... The image has a pixel width of 'width' and a pixel height of 'height'. The set of contours extracted using cv2.findContours is the extracted contour set. , of which contour It is by A set consisting of pixels, i.e. = , For outline The x and y coordinates of the i-th point.

[0057] S6. Perform multi-level filtering on the extracted contour set to remove noise and invalid contours.

[0058] In this embodiment, the boundary contours of all acquired continuous regions are subjected to dual validity screening: first, threshold screening based on area (≥100 pixels); second, grouping and merging based on the same horizontal line (vertical coordinate difference <10 pixels), thereby integrating multiple contour segments belonging to the same crack into a set of points to be fitted, specifically: (a) Area-based threshold filtering: Calculate the area of ​​each contour (using OpenCV's `contourarea` function, which calculates the signed area based on Green's theorem and takes the absolute value), retain only valid contours with an area greater than or equal to 100 pixels, and treat pixels with areas that are too small as noise, thus defining a set of valid contours. ,in, The area is calculated using OpenCV's contour area formula, where the area threshold can be adjusted according to the image resolution: if the vertical resolution is 10 pixels per inch, then 100 pixels corresponds to an actual physical area of ​​about 10 square inches, which can filter out tiny white spots caused by rock debris or mud bubbles.

[0059] (b) Grouping and merging based on the same horizontal line: Determining if two contours are on the same horizontal line: If the following conditions are met, then it is determined to be "the same horizontal line": Where T=10 is the threshold. , The ordinates are the center coordinates of the two contours.

[0060] Grouping: Let the initial grouping set be... Execution: Check the collection Does the group already exist? and The first outline The condition of "same horizontal line" is met. If such a group exists. Then Join this group ,Right now If it does not exist, create a new group. This results in the final set of contour groups, with each group corresponding to a candidate crack.

[0061] S7. Using a sinusoidal function model, combined with the physical constraints of periodic extension rules and amplitude constraints, a hybrid curve fitting is performed to generate a sinusoidal crack curve with a complete cycle.

[0062] In this embodiment, a periodic extension rule is set to extend fractures that are less than one period into a complete period. Electro-imaging images typically cover a 360-degree area around the well, i.e., horizontally. This corresponds to a complete circle. If a crack extends through the entire well circumference, its sine curve should exactly display one complete cycle. However, due to image boundary truncation or cracks developing only locally, the actual extracted contour points may only cover a portion of the x-range. This embodiment forces periodic extension, fixing the period T = width and the angular frequency B = ... If the fracture profile does not fully represent a cycle in the image, a cycle extension is performed based on the image width. That is, the period T is set to equal the image width, and the complete cycle curve is supplemented in the extrapolated part of the image by the periodicity of the sine function, ensuring that the fitted curve covers the entire logging image range.

[0063] In this embodiment, constraint control is implemented to ensure that the coordinate points of the fitted sinusoidal curve seam do not exceed the image boundary and are limited to one complete cycle of the fitted curve in the image, through physical constraints. (a) Amplitude limit is A≤200: Limiting the amplitude A ensures that the angle of the fitted curve is not too large.

[0064] (b) The period is limited to a complete period within 0 to width pixels. If it is less than one period, it is extended to a complete period. The fixed calculation of angular frequency B is as follows: ; Since B is fixed, when x has only a portion of pixels, the curve will naturally extend according to a sine law, completing the full periodic shape (for example, returning from the trough to the peak, and then back to the starting phase), thus achieving periodic extension.

[0065] In this embodiment, the crack curve is fitted using a sinusoidal curve model. Based on the periodic characteristics of cracks in electro-optical imaging images, a sinusoidal function Y=A·sin(B·x+C)+D is used for fitting, where A is the amplitude (reflecting the degree of crack opening and closing), B is the angular frequency (reflecting the period), C is the initial phase, and D is the baseline offset (reflecting the average crack depth). This ensures that as x changes from 0 to the image width, ... This corresponds exactly to one complete cycle of the sine function. By substituting the previously obtained contour points as x, we can fit the curve.

[0066] S8. Curve drawing and output.

[0067] In this embodiment, the fitting parameters (amplitude, phase, baseline, etc.) of the fitted complete periodic sine curve are saved to a fracture parameter table (including well depth range, amplitude, phase, baseline, dip angle, strike, etc.). On the electrical imaging logging image (grayscale or color image), a complete sine curve is drawn in red or a highlight color according to the above parameters, with a line width of 2 pixels. Simultaneously, the fracture type (gdf / gzf / yz) is labeled, and the image file (PNG / JPG) showing the sine curve drawn on the electrical imaging logging image, the fracture parameter table, the binarized segmentation results, and the intermediate results after morphological processing are output to obtain the final recognition effect, such as... Figure 7 The image shows the final recognition result of high-resistance seams. You can see that the light-colored areas are high-resistance seams, which have all been recognized.

[0068] The specific application of the electro-imaging sinusoidal crack identification method based on the hybrid curve fitting algorithm of the present invention will be described in detail below through specific embodiments.

[0069] Example 1: This example uses a high-resistance seam as an example. The image size in this example is expanded to 420 pixels, specifically: 1) Use electrical imaging logging images as input, such as... Figure 2 As shown.

[0070] 2) Input the electrical imaging logging image into the trained image segmentation model to obtain a binarized image: such as Figure 3 As shown, the white pixels correspond to the high-resistivity seam GZF ( Figure 2 The light-colored sinusoidal lines in the image represent the background color, while the black in the binarized image represents the background color. Figure 3 The results of the segmentation of four high-resistance seam curves were obtained, but it can be found that the uppermost curve has a break in the middle and is not completely segmented.

[0071] 3) Employ a multi-contour extraction method to extract the contour of each high-resistance seam: such as... Figure 4 As shown, different colors are used to display the original contour and fitted curve of each group of images in the lower left corner. Contour extraction can be seen from the figure. The points of the upper and lower boundaries of the contour are extracted and saved to prepare for the next step of curve fitting.

[0072] 4) Validity screening: For isolated noise points in the segmentation results, points with a screening area less than 100 are not considered for the fitted curve. If the segmentation results are good, this step is unnecessary. Figure 3 The mid-segmentation has no obvious isolated points scattered in other locations of the image.

[0073] 5) Crack curve fitting: such as Figure 5 As shown, a sine curve model is constructed, and the large number of previously extracted contour points are fitted using the sine function Y=A·sin(B·x+C)+D.

[0074] 6) Constraint control and periodic extension: such as Figure 6 The left image shows the crack fitting plot without constraints, where some of the fitted curves can be seen outside the image width range, such as... Figure 6 The right figure shows the effect after using constraint control and periodic extension. The period is limited to the image width, and the periodically extended curve, when x changes from 0 to the image width, B∈(0- This corresponds exactly to one complete cycle of the sine function.

[0075] 7) The process of plotting the fitted curve: Input the original image and the crack parameter table, and plot the curve according to the fitting parameters in the table to obtain the final identification result, such as... Figure 7 The final recognition result of the high-resistance seam shown can be seen in the light-colored area, which is the high-resistance seam and has been fully recognized.

[0076] Example 2: This example uses a feather-shaped induced suture as an example. The image size in this example is 360 pixels wide. The specific process is as follows: 1) such as Figure 8 As shown on the left, the original electrical imaging logging image of the feather-shaped induced fracture is used as input; after segmentation calculation by the image segmentation model, the segmented feather-shaped fracture region image is represented in white, as shown below. Figure 8 As shown on the right, the segmentation results of multiple feather-shaped induced suture curves are obtained here.

[0077] 2) Perform multi-profile extraction and validity screening, such as... Figure 9As shown, contour extraction is performed on each feather-shaped slit curve. The red point set represents the contour boundary point set of the pixel region of each feather-shaped slit after segmentation. Valid contours with an area greater than or equal to 100 pixels are retained, and multiple contours are grouped based on the same horizontal line. For isolated noise points appearing in the segmentation results, points with an area less than 100 are excluded from the fitting curve.

[0078] 3) Crack curve fitting and constraint control, periodic extension: A sine function is used to fit and calculate multiple previously extracted contour points to generate a sine curve for a single complete cycle. Meanwhile, since the amplitude of feather-shaped induced cracks is usually smaller than that of high-guide and high-resistance cracks, tests showed that setting the amplitude selection rule to A≤100 resulted in better identification performance.

[0079] 4) The process of plotting the fitted curve: Input Figure 8 The original image on the left and the crack parameter table are used to plot a curve based on the fitting parameters in the table, as shown below. Figure 9 The final identification and fitting result is shown on the right. The red curve represents the identification and fitting result of the feather-shaped induced suture.

[0080] In summary, the sinusoidal fracture identification method based on a hybrid curve fitting algorithm provided by this invention adopts a technical approach of "deep learning segmentation + morphological processing + multi-level contour screening + hybrid curve fitting". It uses an image segmentation model to identify and segment the sinusoidal fracture pixel region from the image background, and then uses a hybrid curve fitting algorithm to further refine the segmented region and fit it into a complete periodic sinusoidal curve. The hybrid curve fitting algorithm includes multi-contour extraction, periodic extension rules, validity screening, and physical constraints. Therefore, this invention can automatically, efficiently, and robustly identify high-conductivity fractures, high-resistance fractures, and feather-induced fractures, and accurately fit a complete periodic sinusoidal curve, providing a reliable basis for fracture occurrence calculation and reservoir evaluation.

[0081] Example 2: Following the above example 1 which provided a method for identifying sinusoidal cracks in electro-imaging based on a hybrid curve fitting algorithm, this example provides a device for identifying sinusoidal cracks in electro-imaging based on a hybrid curve fitting algorithm. The device provided in this example can implement the method for identifying sinusoidal cracks in electro-imaging based on a hybrid curve fitting algorithm as described in Example 1. This device can be implemented through software, hardware, or a combination of both. For ease of description, this example is described by dividing the functionality into various units. Of course, in implementation, the functions of each unit can be implemented in one or more software and / or hardware components. For example, the device may include integrated or separate functional modules or units to execute the corresponding steps in the methods of Example 1. Since the device in this example is basically similar to the method example, the description process of this example is relatively simple. Relevant details can be found in the description of Example 1. The example of the device for identifying sinusoidal cracks in electro-imaging based on a hybrid curve fitting algorithm provided by this invention is merely illustrative.

[0082] Specifically, the present invention also provides an electro-imaging sinusoidal crack identification device based on a hybrid curve fitting algorithm, comprising: The data acquisition unit is configured for structured data acquisition and cleaning; The data processing unit is configured to convert the cleaned structured data into electrical imaging logging images, wherein the electrical imaging logging images include high-conductivity fractures, high-resistance fractures and / or feather-shaped induced fractures; The image segmentation unit is configured to input electrical imaging logging images into a pre-trained image segmentation model and output a binary image, wherein the pixel value of the fracture region in the binary image is 255 and the background is 0; The image binarization unit is configured to perform morphological operations on the binary image; The contour acquisition unit is configured to find the boundaries of all consecutive non-zero pixel regions in the contour of the binary image after morphological operations, and obtain a contour set. The contour filtering unit is configured to perform multi-level filtering on the contour set to remove noise and invalid contours. The curve fitting unit is configured to use a sinusoidal function model, combined with physical constraints of periodic extension rules and amplitude constraints, to perform mixed curve fitting on the screened contour set, generating a sinusoidal crack curve with a complete cycle.

[0083] Example 3: This example provides an electronic device corresponding to the electro-imaging sinusoidal crack identification method based on hybrid curve fitting algorithm provided in Example 1. The electronic device can be an electronic device for the client, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Example 1.

[0084] like Figure 10As shown, the electronic device includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus to enable communication between them. The memory stores a computer program that can run on the processor. When the processor runs the computer program, it executes the method of Embodiment 1. The implementation principle and technical effects are similar to those of Embodiment 1, and will not be repeated here. Those skilled in the art will understand that... Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computing device on which the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0085] In a preferred embodiment, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), and optical discs.

[0086] In a preferred embodiment, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and is not limited thereto.

[0087] Example 4: This example provides a computer-readable storage medium for storing one or more programs, the one or more programs including computer instructions, which, when executed by a computer, cause the computer to perform the method provided in Example 1 above.

[0088] In a preferred embodiment, the computer-readable storage medium may be a tangible device for holding and storing instructions executable, such as, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. The computer-readable storage medium stores computer program instructions that cause a computer to perform the method provided in Embodiment 1 above.

[0089] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (devices), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0090] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0091] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0092] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In the description of this specification, the terms "a preferred embodiment," "furthermore," "specifically," "in this embodiment," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the embodiments in this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0093] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; 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 sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm, characterized in that, include: Structured data acquisition and cleaning; The cleaned structured data is converted into electrical imaging logging images, which include high-conductivity fractures, high-resistance fractures, and / or feather-shaped induced fractures. The electrical imaging logging image is input into a pre-trained image segmentation model, which outputs a binary image. In the binary image, the pixel value of the fracture region is 255 and the background is 0. Perform morphological operations on binary images; Find the boundaries of all consecutive non-zero pixel regions in the binary image contour after morphological operations to obtain the contour set; Perform multi-level filtering on the contour set to remove noise and invalid contours; A sinusoidal function model is used, combined with the physical constraints of periodic extension rules and amplitude constraints, to perform mixed curve fitting on the screened contour set and generate a sinusoidal crack curve with a complete cycle. The sinusoidal fracture curve and the electrical imaging logging image are superimposed to complete the fracture identification.

2. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 1, characterized in that, Structured data acquisition and cleaning, including: Data was simultaneously acquired using a downhole high-resolution multi-plate array micro-resistivity imager, including resistivity matrix, natural gamma curve, well inclination and azimuth, mud resistivity, and well temperature data. The collected data is cleaned and corrected; The cleaned and corrected data are stored in depth order as a format file as a standardized base dataset.

3. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 2, characterized in that, The cleaned structured data is converted into electrical imaging logging images, including: Convert the resistivity matrix into an electrical imaging logging image; Based on the well inclination azimuth information, the electrical imaging logging images are azimuth corrected so that the top of the electrical imaging logging images points to geographic north or magnetic north. Among them, the fracture characteristics in the electrical imaging logging images are as follows: high-conductivity fractures appear as dark or black linear / curved features; high-resistance fractures appear as light or white linear features; and feather-shaped induced fractures appear as complex curved shapes resembling tree branches or feathers.

4. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 1, characterized in that, The image segmentation model is trained using an improved YOLOv11seg deep learning segmentation model. This improved model integrates a C3K2 enhanced feature extraction module, a SPFF multi-scale fusion module, an adaptive anchor box mechanism, and an EIOU loss function. The training process for the image segmentation model is as follows: Several electro-imaging image samples of labeled crack regions were collected as a training set. The labeled crack regions included high-conductivity cracks, high-resistance cracks, and feather-induced cracks. The weights pre-trained on the COCO dataset were used as initial parameters for fine-tuning through transfer learning. The optimizer used was AdamW with an initial learning rate of 0.001, momentum of 0.9, weight decay of 0.0005, batch size of 8, and training for 50 epochs. After training, the image segmentation model was obtained.

5. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 1, characterized in that, Morphological operations are performed on the binary image, including opening and closing operations. Opening is used to remove isolated foreground noise and cut off small spiky connections; closing is used to fill tiny holes inside cracks and connect broken segments caused by discontinuity in segmentation. The execution order and number of iterations are as follows: first, open is performed once, then closing is performed once. In areas with low foreground probability, the number of dilation operations in closing is reduced to avoid spurious connections, resulting in an optimized binary image.

6. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 1, characterized in that, The contour set undergoes multi-level filtering to remove noise and invalid contours. Specifically, all acquired continuous region contours are subjected to dual validity filtering: first, area-based threshold filtering; second, grouping and merging based on the same horizontal line, thereby integrating multiple contour segments belonging to the same crack into a single set of points to be fitted. (a) Area-based threshold filtering: Calculate the area of ​​each contour and retain only valid contours with an area greater than or equal to a set number of pixels; (b) Grouping and merging based on the same horizontal line: Determining if two contours are on the same horizontal line: If the following conditions are met, then it is determined to be "the same horizontal line": Where T is the threshold. , The ordinate of the center of the two contours; Grouping: Let the initial grouping set be... Execution: Check the collection Does the group already exist? and The first outline If a group satisfies the condition of "same horizontal line", then such a group exists. Then Join this group ,Right now If it does not exist, create a new group. This results in the final set of contour groups, with each group corresponding to a candidate crack.

7. The method for identifying sinusoidal cracks in electrical imaging based on a hybrid curve fitting algorithm according to claim 1, characterized in that, A sinusoidal function model is used, combined with physical constraints such as periodic extension rules and amplitude constraints, to perform hybrid curve fitting on the screened contour set, generating a complete-cycle sinusoidal crack curve, including: Periodic continuation rule: Forced periodic continuation, fixed period T = width, angular frequency B = If the fracture profile does not fully represent a cycle in the image, the cycle is extended based on the image width. That is, the cycle T is set to equal the image width, and the complete cycle curve is supplemented in the extrapolated part of the image by the periodicity of the sine function, so as to ensure that the fitted curve covers the entire logging image range. Physical constraints of amplitude constraint: (a) Amplitude is limited to A≤ preset value; (b) Period is limited to a complete period within 0~width pixels, and if it is less than a period, it is extended to a period. Crack curve fitting: Based on the periodic characteristics of cracks in electro-imaging images, a sine function Y=A·sin(B·x+C)+D is used for curve fitting, where A is the amplitude, B is the angular frequency, C is the initial phase, and D is the baseline offset.

8. A sinusoidal crack identification device for electrical imaging based on a hybrid curve fitting algorithm, characterized in that, include: The data acquisition unit is configured for structured data acquisition and cleaning; The data processing unit is configured to convert the cleaned structured data into electrical imaging logging images, wherein the electrical imaging logging images include high-conductivity fractures, high-resistance fractures and / or feather-shaped induced fractures; The image segmentation unit is configured to input electrical imaging logging images into a pre-trained image segmentation model and output a binary image, wherein the pixel value of the fracture region in the binary image is 255 and the background is 0; The image binarization unit is configured to perform morphological operations on the binary image; The contour acquisition unit is configured to find the boundaries of all consecutive non-zero pixel regions in the contour of the binary image after morphological operations, and obtain a contour set. The contour filtering unit is configured to perform multi-level filtering on the contour set to remove noise and invalid contours. The curve fitting unit is configured to use a sinusoidal function model, combined with physical constraints of periodic extension rules and amplitude constraints, to perform mixed curve fitting on the screened contour set, generating a sinusoidal crack curve of the complete cycle. The fracture identification unit is configured to overlay sinusoidal fracture curves and electrical imaging logging images to identify fractures.

9. An electronic device, characterized in that, include: At least one processor; And a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to enable the processor to perform the method according to any one of claims 1-7.

10. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include computer instructions for causing a computer to perform the method according to any one of claims 1-7.

Images