Porous medium CT image segmentation method, system and device and medium

Abstract

The invention relates to a porous medium CT image segmentation method, system and device and a medium, and the method comprises the steps: obtaining a rock core sample, and scanning the rock core sample through a CT device to obtain a CT image of the rock core sample; the method comprises the following steps: preprocessing a CT image of a rock core sample, and extracting a main part of gray level distribution in the CT image as an image to be analyzed; performing multi-distribution fitting on the to-be-analyzed image to obtain a multi-distribution fitting image; and performing intelligent threshold segmentation on the multi-distribution fitting image to obtain an image segmentation result containing different pore filler boundaries. According to the method, through gray feature extraction, multi-distribution fitting, intelligent threshold segmentation and result optimization, precise segmentation of pores containing different pore fillers in the rock is realized, the precision and reliability of pore filler analysis are remarkably improved, and the method can be widely applied to the technical field of intelligent identification of porous medium components.

Description

Technical Field

[0001] This invention belongs to the field of intelligent identification technology of porous media components, specifically relating to a method, system, device and medium for segmenting porous media CT images considering heel effect and pore filling material, used to solve the problem of intelligent segmentation of the skeleton and pore filling material in rock CT images under heel effect. Background Technology

[0002] In the field of porous media composition analysis, accurate segmentation of the framework and pore fillers using CT images is one of the core tasks. Currently, CT image scanning, such as the commonly used third-generation CT (rotation-rotation) and cone-beam CT, suffers from heel effect during scanning, causing uneven grayscale between different regions and severely affecting segmentation accuracy. Simply removing the areas with severe heel effect leads to a smaller dataset, making it difficult to fully understand the proportion and characteristics of the rock framework, resulting in significant discrepancies between simulation and experimental results based on the rock framework. Furthermore, the pore structure of the rock samples to be analyzed is complex and often contains various pore fillers (such as oil, water, and gas). The distribution of different pore fillers within the pores and the boundaries between the pores themselves are often unclear, especially since the grayscale value of the oil phase after CT scanning is not significantly different from that of the pores. This makes it extremely difficult to successfully separate the pores from the oil phase when it is used as a pore filler.

[0003] Existing traditional adaptive methods perform poorly in handling cases of uneven grayscale. The mathematical models constructed from CT image scans are crucial for subsequent physical simulations, requiring high accuracy in pore filler segmentation. Global thresholding suffers from poor results due to the heel effect and difficulty in identifying boundaries, failing to adapt to grayscale variations. Current adaptive thresholding methods, such as local mean, Otsu's algorithm, and maximum inter-class difference, typically yield unique thresholds for each image. However, their accuracy is compromised in multi-class segmentation, especially when thresholds are difficult to define. Furthermore, the obtained thresholds are hard to correlate with components, resulting in poor interpretability. The performance is also unsatisfactory when pore filler and pore boundaries are blurred.

[0004] Especially when segmenting pores containing different pore fillers, the blurred boundaries between the fillers and pores, coupled with the low grayscale difference due to the heel effect, make accurate differentiation of different pore fillers extremely difficult. This is particularly true for oil-bearing pores, where the grayscale difference between the oil phase and the pores is small, making it difficult to distinguish them using conventional adaptive methods. However, clearly defining the distribution of the oil phase in oil-bearing pores is crucial, as it plays a decisive role in subsequent scientific calculations and engineering implementations such as geological assessment, rock mechanics parameter calculations, and oil and gas extraction. Therefore, there is an urgent need for a method that can effectively solve the above problems and achieve accurate segmentation of rocks containing different pore fillers. Summary of the Invention

[0005] To address the aforementioned problems, the present invention aims to provide a method, system, device, and medium for segmenting porous media CT images. By extracting grayscale features, fitting multiple distributions, intelligent thresholding, and optimizing results, it overcomes the limitations of traditional adaptive methods and global thresholding in processing such images. This enables accurate segmentation of pores containing different pore-filling materials, providing reliable image analysis data support for subsequent oil and gas development and other engineering operations, and improving the accuracy and effectiveness of engineering implementation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for segmenting porous media CT images, comprising: Core samples were obtained and scanned using CT equipment to obtain CT images of the core samples. The CT images of the core samples were preprocessed to extract the main part of the gray-scale distribution in the CT images as the images to be analyzed. The image to be analyzed is fitted with a multi-distribution model to obtain a multi-distribution fitted image. Intelligent thresholding segmentation is performed on multi-distribution fitted images to obtain image segmentation results containing the boundaries of different pore filling materials.

[0007] In a second aspect, the present invention provides a porous media CT image segmentation system, comprising: The image acquisition module is used to acquire core samples and scan the core samples using CT equipment to obtain CT images of the core samples; The data preprocessing module is used to preprocess the CT image data of the core sample and use the main part of the grayscale distribution in the processed CT image as the image to be analyzed. The multi-distribution fitting module is used to perform multi-distribution fitting on the image to be analyzed, and obtain a multi-distribution fitted image. The intelligent thresholding segmentation unit is used to perform intelligent thresholding segmentation on multi-distribution fitted images to obtain image segmentation results containing the boundaries of different pore filling materials.

[0008] Thirdly, the present invention provides a computer-readable storage medium storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the porous media CT image segmentation method.

[0009] Fourthly, the present invention provides a computing device comprising: one or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs comprising instructions for performing any of the methods in the porous media CT image segmentation method.

[0010] The present invention has the following advantages due to the adoption of the above technical solutions: (1) This invention can effectively solve the problem of component extraction under the action of heel effect and conical light coupling by gray-scale feature extraction, multi-distribution fitting, intelligent threshold segmentation and result optimization. That is, it can still extract image components when the gray-scale is uneven between images, and can accurately segment pores and asphalt with similar gray-scale values, which is very close to the actual results.

[0011] (2) By comparing the porosity extracted by the present invention with that extracted by the mainstream adaptive threshold scheme, the present invention demonstrates that the results of the present invention are better than those of commonly used algorithms in handling this case.

[0012] (3) This invention improves the stability and accuracy of segmentation by improving the traditional GMM method and combining a preprocessing scheme, and determines the classification result by combining the fractal dimension.

[0013] (4) This invention uses two novel threshold classification schemes based on probability distribution applicable to different situations, combined with other requirements and known information, to further optimize the misclassification rate in soft classification. This allows for a more intuitive and effective understanding of the error of the segmentation scheme, and facilitates researchers in controlling and measuring the accuracy of the experiment.

[0014] Therefore, this invention can be widely applied in the field of intelligent identification technology for porous media components. Attached Figure Description

[0015] 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 porous media CT image segmentation method provided in the embodiments of the present invention; Figure 2 This is a schematic diagram illustrating the different grayscale levels caused by the heel effect and conical optical coupling provided in an embodiment of the present invention; Figures 3a to 3d This is a schematic diagram illustrating the entire process of determining the threshold through two fitting steps provided in this embodiment of the invention, wherein... Figure 3a The porosity threshold determined by a single fitting. Figure 3bThe confidence region is the fractal dimension corresponding to different peaks in a single fitting. Figure 3c The pore threshold is determined by the quadratic fitting. Figure 3d The confidence region for the fractal dimension corresponding to different peaks in the second-order fitting; Figures 4a-4c These are the fractal dimensions of the extraction results at different resolutions provided in the embodiments of the present invention, wherein, Figure 4a For high-resolution fractal dimensions of each component, Figure 4b For low-resolution fractal dimensions of each component, Figure 4c The fractal dimensions of each component at low resolution after removing the first 25% of the data; Figure 5 These are the extraction results of CT images with different average gray levels provided in the embodiments of the present invention; Figure 6 This is a comparison chart of the asphalt ratio and porosity at different resolutions and the measured values ​​provided in the embodiments of the present invention; Figure 7 This is the distribution of pore identification accuracy under different threshold segmentation schemes at high resolution provided in the embodiments of the present invention; Figure 8 These are high-precision extraction effect diagrams using different algorithms provided in the embodiments of the present invention; Figure 9 These are low-precision extraction effect diagrams of different algorithms provided in the embodiments of the present invention; Figure 10 The diagram shows the effect of three-phase pore segmentation using a conventional adaptive segmentation scheme and the method of the present invention, as provided in the embodiments of the present invention. Detailed Implementation

[0016] 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0017] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0018] To address issues such as uneven grayscale caused by the heel effect, complex rock pore structure, and blurred boundaries between pores and different pore-filling materials (e.g., oil, water), some embodiments of this invention provide a porous media CT image segmentation method. Through grayscale feature extraction, multi-distribution fitting, intelligent threshold segmentation, and result optimization, this method fully utilizes the image's grayscale distribution characteristics, effectively overcomes the interference of the heel effect, and accurately identifies pores in rocks containing different pore-filling materials. This meets the requirements of high-precision threshold segmentation for mathematical models of CT image construction, laying a solid foundation for rock physics simulation and engineering applications, and achieving accurate segmentation of rock pores and their filling materials.

[0019] Correspondingly, in other embodiments of the present invention, a porous medium CT image segmentation system, device, and medium are provided.

[0020] Example 1 like Figure 1 As shown, the present invention provides a method for segmenting porous media CT images, which includes the following steps: S1. Obtain core samples and scan them using CT equipment to obtain CT images of the core samples.

[0021] S2. Preprocess the CT images of the core samples and extract the main part of the gray-scale distribution in the CT images as the images to be analyzed.

[0022] For a set of rock CT images, the following factors mainly affect segmentation in porous media CT images: gray-level differences within a single image, primarily caused by various artifacts (such as ring artifacts and hardening artifacts); gray-level differences between different images, mainly caused by the coupling of heel effect and cone light effect, such as... Figure 2 As shown, image noise is mainly caused by a variety of noises, such as quantum noise, scattering noise, and detector noise.

[0023] Therefore, this embodiment requires preprocessing of the acquired CT images, including the following steps: S21. For the acquired single CT image, a region is cropped at a preset position in the original CT image to obtain the CT image of interest, so as to reduce the influence of ring artifacts and hardening artifacts.

[0024] S22. Perform grayscale mapping and main part extraction on the CT image of interest, and use the main part of the grayscale distribution in the extracted CT image as the image to be analyzed.

[0025] To fully utilize grayscale features, this embodiment requires in-depth processing of the grayscale data of rock CT images, focusing on extracting the main parts of the grayscale distribution to effectively filter out abnormal grayscale fluctuations caused by the heel effect and interference data caused by the rock matrix, ensuring the validity and representativeness of subsequent analysis data, and focusing on preserving the grayscale features of pores and internal pore filling materials; at the same time, to address noise issues, this embodiment uses a preset filtering method to reduce image noise.

[0026] When extracting the main part of the grayscale distribution, the process is as follows: First, the gray values ​​of the CT images of interest are uniformly mapped to 8-bit grayscale to obtain a grayscale histogram; Secondly, based on the preset grayscale threshold, the main part of the grayscale histogram is extracted as the image to be analyzed.

[0027] After extracting the main part based on the grayscale threshold, the grayscale distribution of the CT image. Represented as:

[0028] in, This indicates that the grayscale value is lower than The number of points, M represents the total number of points in the entire graph. These are preset hyperparameters. This represents the original grayscale distribution in the CT image of interest, also known as the probability density function.

[0029] S3. Perform multi-distribution fitting on the image to be analyzed to obtain a multi-distribution fitted image.

[0030] Specifically, it includes the following steps: S31. Analyze the grayscale distribution of the image to be analyzed to determine the initial number of components, that is, the number of different components contained in the core sample.

[0031] In this embodiment, based on the principle of mixed probability distribution fitting, the image to be analyzed is regarded as a superposition of multiple Gaussian distributions or multiple Cauchy distributions. Empirically, for porous media such as rock, the lowest gray level is always found in pores, and the second lowest gray level is always found in asphalt (if present). That is, different peaks belong to different components. Based on this, the peak shapes of the main gray-scale distribution of the image to be analyzed are used to determine the different components, which are then used as the initial number of components to achieve refined modeling of the gray-scale characteristics of pores and different pore-filling materials.

[0032] S32. Based on the determined initial number of components, the improved GMM algorithm is used to perform multi-distribution fitting on the image to be analyzed, and the average residual corresponding to the multi-distribution fitting result under the current number of components is determined.

[0033] Specifically, it includes the following steps: S321. Based on the characteristics of the rock sample, determine the range of parameter values ​​for each component, that is, determine the upper and lower limits for each component. S322. Based on the determined range of parameter values, configure the hyperparameters of each component for the current iteration number, wherein the hyperparameters include the distribution parameters of each component and the corresponding weights. In this embodiment, when performing multi-distribution fitting on the image to be analyzed, either Cauchy distribution or Gaussian distribution can be used. When using Cauchy distribution for fitting, because the tail of the Cauchy distribution is thicker, it better matches the characteristics of the component distribution and reduces the normality requirement of the data. This allows the GMM algorithm to no longer rely solely on the light-tailed distribution with Gaussian distribution as the leader, thus improving the outlier resistance of the mixture model.

[0034] When using the Cauchy distribution for multi-distribution fitting, the calculation formula is expressed as:

[0035] In the formula, The grayscale distribution of the image, The number of Cauchy components. For the first The weights of each Cauchy component For the first Distribution parameters of the Cauchy components, For the first Position parameters of each Cauchy component For the first The scale parameter of each Cauchy component; When using a Gaussian distribution for fitting, the calculation formula is expressed as:

[0036] In the formula, To analyze the grayscale distribution of the image, The number of Gaussian components. For the first The weights of the Gaussian components, For the first Distribution parameters of Gaussian components, For the first The mean of the Gaussian components, For the first The variance of each Gaussian component.

[0037] Among them, the hyperparameters corresponding to the Cauchy distribution or hyperparameters corresponding to a Gaussian distribution The parameters need to be estimated iteratively using the Expectation-Maximization (EM) algorithm. However, the conventional EM algorithm does not impose specific restrictions on the range of these three parameters, so it is prone to getting trapped in local optima during fitting. A typical manifestation of this is the appearance of a Gaussian function with a large variance as the overall fit.

[0038] To prevent the model from getting trapped in local optima, this invention incorporates the following variance constraint during the iteration process of the EM algorithm:

[0039] In the formula, For the first During the nth iteration The variance of the Gaussian component or the first Gaussian component The scale parameter of each Cauchy component This is the lower limit of the variance or scaling parameter. This represents the upper limit of the variance or scale parameter. For the first The grayscale value of a pixel in a Gaussian or Cauchy component; For the first During the nth iteration The weights of Gaussian or Cauchy components; For the first During the nth iteration The mean of the nth Gaussian component or the 1st Gaussian component The scale parameter of each Cauchy component.

[0040] S323. Perform multi-distribution fitting based on the determined hyperparameters, and calculate the average residual corresponding to the multi-distribution fitting result at the current iteration number; S324. Determine whether the average residual corresponding to the multi-distribution fitting result under the current iteration number is less than the preset threshold. If it is, stop the iteration and obtain the optimal hyperparameter corresponding to the current number of components. Otherwise, update the hyperparameter according to the preset step size and return to step S323 until the optimal hyperparameter is obtained. S325. Take the average residual corresponding to the multi-distribution fitting result of the optimal hyperparameter as the average residual corresponding to the multi-distribution fitting result under the current number of components.

[0041] S33. Determine whether the average residual corresponding to the multi-distribution fitting result under the current number of components is less than the preset threshold. If so, stop the iteration. Otherwise, update the number of components according to the preset step size and return to step S32 to perform multi-distribution fitting until the convergence condition is met, obtain the optimal number of components and its corresponding hyperparameters, and complete the multi-distribution fitting.

[0042] S4. Perform intelligent thresholding on the multi-distribution fitted image to obtain image segmentation results containing the boundaries of different pore filling materials.

[0043] Specifically, it includes the following steps: S41. Based on Bayesian theory or the likelihood ratio algorithm based on hypothesis testing, classify all pixels in the multi-distribution fitted image to obtain the preliminary boundaries of different pore filling components.

[0044] After completing the multi-distribution fitting, the result only represents the probability that each pixel in the image belongs to a specified component. To determine which component each pixel belongs to, this embodiment uses an algorithm based on Bayesian theory or a likelihood ratio based on hypothesis testing for accurate classification. The difference between these two approaches is that if the proportion of each component has already been estimated, using the Bayesian method will significantly improve the segmentation accuracy. However, if it is desirable to protect a certain component and ensure that the segmentation error probability of that component does not exceed a certain value, then a likelihood ratio based on hypothesis testing should be used for segmentation.

[0045] Specifically, the classification process based on Bayesian theory is as follows: First, the probability that each pixel in the multi-distribution fitted image is misclassified as a different pore-filling component is calculated.

[0046] Among them, the gray value of a certain pixel Mistakenly identified as a component of a certain pore filler The probability is:

[0047] In the formula, For a certain pore-filling component Estimated volume percentage For normalization parameters, grayscale The point belongs to the component The probability of.

[0048] Secondly, based on the probability value of each pixel being misclassified as a component, all pixels in the multi-distribution fitted image are classified to obtain an image containing the boundaries of different pore filling materials.

[0049] In this embodiment, it is only necessary to find the pixel. To minimize the possibility of misclassification when a pixel is classified into a particular component, the classification of pixels in a multi-distribution fitted image is performed, i.e., if...

[0050] Then point Belongs to components .

[0051] The classification process based on likelihood ratios in hypothesis testing is as follows: First, calculate the log-likelihood ratio of the probability that each pixel in the multi-distribution fitted image belongs to a specified pore-filling component and other pore-filling components.

[0052] In this embodiment, it is assumed that the component to be protected is Its corresponding Gaussian component is The sum of the remaining Gaussian components is denoted as For the grayscale value of each pixel This point belongs to the component The log-likelihood ratio of the probability of belonging to one group to the probability of belonging to another group is expressed as:

[0053] when At that time, explain the point. Belongs to components The probability is higher; conversely, when When, explain Belongs to components The possibility is relatively small.

[0054] Then, based on the given threshold and the calculated log-likelihood ratio, each pixel is assigned to the corresponding pore-filling component, thus completing the classification.

[0055] Based on given segmentation components Maximum acceptable missegmentation probability This can be confirmed using the following formula. Components:

[0056] in, This represents the probability of missegmentation. When this probability is less than the acceptable probability of missegmentation, the point can be accepted as belonging to the component. Conversely, this point cannot be considered a component. .

[0057] S42. Perform morphological inspection on the preliminary boundaries of the different pore filling components obtained in step S41, and obtain the final binarized image containing the boundaries of different pore filling components based on the inspection results.

[0058] Specific classification results can be obtained through Bayesian or hypothesis testing-based likelihood ratios. To ensure that each function corresponds to only one component, morphological tests are also required for the components corresponding to different functions.

[0059] In this embodiment, the results of different classifications are counted multiple times using the box counting method to obtain the confidence region of the fractal dimension. If the confidence regions of the fractal dimension of different pore-filling components do not overlap after segmentation, it can be said that the segmentation result of each function corresponding to one component is correct. Otherwise, it is considered that the overlapping peaks of the confidence regions of the fractal dimension should correspond to the same substance, and segmentation is performed according to this result.

[0060] S43. Optimize the binarized image containing the boundaries of different pore filling materials to generate high-precision, standardized image segmentation results.

[0061] In this embodiment, after thresholding the CT image, the binarized image result obtained after thresholding needs to be optimized to eliminate possible segmentation noise and discontinuous regions, generate high-precision and standardized segmentation data and save it, so as to provide high-quality image analysis basic data for subsequent rock physics simulation, engineering scheme design and other purposes, and ensure the accuracy and practicality of the processing results.

[0062] Example 2 This embodiment further introduces the porous media CT image segmentation method proposed in this invention. Based on the oil sand image dataset of the Qigu Formation of the Jurassic in Fengcheng Oilfield, Karamay, Xinjiang, the grayscale range is determined, and the grayscale range of different components is roughly constructed by combining multi-function fitting. Then, the optimal number of fitting functions is obtained by averaging the residuals. The independence of the components represented by each function is verified by fractal dimension. Threshold segmentation is performed using Bayesian judgment or hypothesis testing based on likelihood ratio. The segmentation results are compared with common adaptive threshold segmentation results and experimental results to determine the accuracy of the segmentation.

[0063] In this embodiment, the core samples used were from the Qigu Formation of the Jurassic system in the Fengcheng Oilfield of Karamay, Xinjiang. The oil sands in this formation are terrestrial braided fluvial deposits and are weakly consolidated sediments. Their main components are framework sand, bituminous material, and clay minerals. The mineral components of the particles are mainly quartz and feldspar, with a high mud content, poor cementation, loose structure, and poor bedding. The burial depth is approximately 300-500m.

[0064] The entire oil sands core sample scanning process utilized the Xradia microxct-200 micrometer CT scanner. This device is widely used for CT image scanning and analysis of various types of cores. Its resolution range is 0.5. Up to 35 Core scanning was performed from top to bottom, with a scanning resolution of 2024×2048 pixels. To compare the recognition performance of the invention at different resolutions, 12 pixels were selected in this embodiment. A total of 2752 high-resolution CT images were used as low-resolution samples and 5 A total of 1306 high-precision CT images were used as high-precision samples, of which 12... The high-resolution image is of a core column with a diameter of 38 mm and a height of 76 mm. The high-precision image is a small cylinder with a diameter of 15mm and a height of 30mm, cut from the middle of the original core column.

[0065] like Figure 2 As shown, the difference in grayscale is caused by the heel effect and cone-shaped light coupling. This embodiment preprocesses the acquired CT images, including region cropping, grayscale mapping, and median filtering. Region cropping is performed to reduce the impact of single-image artifacts. In this embodiment, an 800×800px region in the center of the image is directly cropped. Within the original image size of 2048×2048px, this region can reduce the influence of ring artifacts and hardening artifacts.

[0066] Gray-level mapping and median filtering are used to extract the main gray-level distribution in an image. Gray-level mapping refers to mapping the gray levels of the cropped image to a uniform range of 0-255. For ease of comparison and reading, the gray levels of the CT images used in this embodiment are uniformly mapped to 8-bit gray levels. Median filtering is used to denoise the image, reducing the impact of extremely bright or dark pixels on the stability of subsequent algorithms.

[0067] like Figures 3a to 3d The diagram shows the process of quadratic fitting. The first fitting is as follows: Figure 3a As shown, the fractal dimension confidence region was used to determine that the peak with the lowest mean did not intersect with other peaks, as shown in the figure. Figure 3b As shown, at this point, the pore portion with the lowest grayscale can be extracted and its distribution subtracted. The remaining portion is then fitted a second time, and the result of the second fitting is shown below. Figure 3c As shown, the uniqueness of the peak correspondence was determined and the asphalt was extracted using the fractal dimension confidence region, and the results are as follows. Figure 3d As shown, at this point, the grayscale of the asphalt and particles is difficult to distinguish, and the accuracy rate is lower than when extracting particles, but the difference between the extraction result and the actual result is still small.

[0068] By using the global adaptive segmentation scheme of this embodiment on the captured CT images, although the grayscale range of some images in the dataset is extremely large due to the coupling of the heel effect and cone light, the segmentation scheme of this embodiment still has a high accuracy. The extraction results of pores and asphalt in the two datasets of this embodiment are shown in Table 1.

[0069] A comparison of the overall fractal dimensions of the extraction results is performed, such as... Figures 4a-4cAs shown, the first 25% of the data has too low grayscale at low resolution, resulting in many outliers. After removal, it can be found that, except for a few outliers, the fractal dimensions of each component in each image do not intersect. In practical use, if a complete rock column image is not required, this part can be removed to significantly improve the segmentation accuracy.

[0070] Table 1. Porosity and asphalt percentage obtained at different resolutions.

[0071] As shown in the table, the recognition rate reaches over 80% at different resolutions, with even higher accuracy for pore segmentation, reaching 96.17% at high resolution. As mentioned above, the extraction effect on asphalt is worse than that on pores, but it still maintains a usable level at high resolution. Specific extraction results are as follows... Figure 5 As shown, it can be seen that the present invention can extract the pores and asphalt parts in the figure basically correctly under different average grayscale conditions. The extracted CT image can be used for digital core modeling to carry out further mechanical or geological simulation.

[0072] Figure 6 The results of porosity and bitumen segmentation at different resolutions are shown. The porosity and bitumen percentage identified by this invention generally fluctuate around the measured values. However, due to the low contrast between different components in the lower grayscale region at low resolution, some volumetric effects are severe, leading to an oversensitivity to the grayscale threshold. This results in bitumen and some particles having almost identical grayscale values, ultimately presenting an overestimation of the bitumen percentage. In practical applications, when encountering low grayscale contrast and large fluctuations in the identification threshold, the threshold obtained in this part can be smoothed to suppress the segmentation result fluctuations caused by low contrast. For example, when the requirement for digital core integrity is low and a large amount of data is not needed, this part of the rock image can be removed, which can significantly improve the accuracy of identification.

[0073] Currently, common adaptive algorithms include maximum entropy, factorization, moments, and iterative self-organizing clustering (Isodata). These algorithms exhibit heel effect coupled with conical light effect in segmentation, as shown in... Figure 7 As shown.

[0074] At higher resolutions, the porosity identified by these algorithms still differs significantly from the measured porosity. The figure shows a jump in the proportion of pores. This is because in some images, bright minerals such as pyrite have a large area, which has a greater impact on the grayscale extreme value. In addition, some particles have a low grayscale value, which is similar to asphalt and pores, resulting in the porosity identification result being close to 1.

[0075] like Figure 8 and Figure 9 As shown in the box plot, the differences between each algorithm and the measured values ​​are relatively large. Only the maximum entropy algorithm obtains an IQR corresponding to the pore distribution that is close to the measured porosity, but it is still not penetrated by the measured porosity. That is, the final identification result has a large error with the measured porosity. In contrast, the IQR corresponding to the pore distribution obtained by the algorithm used in this invention is penetrated by the measured porosity and is almost concentrated near the measured porosity. The overall variance is small, indicating that the algorithm proposed in this paper has higher accuracy and better stability.

[0076] Various commonly used adaptive segmentation schemes and the pore segmentation effect of the present invention are as follows: Figure 10 As shown, both the maximum entropy and factorization algorithms performed similar segmentations of the image, failing to completely distinguish between asphalt and particles. The moment algorithm, on the other hand, identified the entire image as particles. This might be because the large proportion of particles in the image resulted in a unimodal grayscale distribution, causing the moment algorithm to fail completely. Furthermore, the iterative self-organizing clustering algorithm does not support three-phase segmentation. On the one hand, the pore segmentation results obtained by this invention align more closely with subjective intuition; other algorithms excessively segmented pores, including asphalt and even some particles. On the other hand, this invention provides relatively complete pore extraction, with almost no discrete gray points. This is beneficial for further mechanical or geological simulations, preventing significant discrepancies between connected pores and the identified pores.

[0077] Example 3 The above-described embodiment 1 provides a porous media CT image segmentation method. Correspondingly, this embodiment provides a porous media CT image segmentation system. The system provided in this embodiment can implement the porous media CT image segmentation method of embodiment 1. The system can be implemented by software, hardware, or a combination of both. For example, the system may include integrated or separate functional modules or units to perform the corresponding steps in the methods of embodiment 1. Since the system in this embodiment is basically similar to the method embodiment, the description process in this embodiment is relatively simple. For relevant details, please refer to the description of embodiment 1. The system embodiment provided in this embodiment is merely illustrative.

[0078] The porous media CT image segmentation system provided in this embodiment includes: The image acquisition module is used to acquire core samples and scan the core samples using CT equipment to obtain CT images of the core samples; The data preprocessing module is used to preprocess the CT image data of the core sample and use the main part of the grayscale distribution in the processed CT image as the image to be analyzed. The multi-distribution fitting module is used to perform multi-distribution fitting on the image to be analyzed, and obtain a multi-distribution fitted image. The intelligent thresholding segmentation unit is used to perform intelligent thresholding segmentation on multi-distribution fitted images to obtain image segmentation results containing the boundaries of different pore filling materials.

[0079] Example 4 This embodiment provides a processing device corresponding to the porous media CT image segmentation method provided in Embodiment 1. The processing device can be a client-side processing device, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Embodiment 1.

[0080] The processing 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 porous media CT image segmentation method provided in Embodiment 1.

[0081] Preferably, the memory may be high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device.

[0082] Preferably, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and there is no limitation herein.

[0083] Example 5 The porous media CT image segmentation method of this embodiment 1 can be specifically implemented as a computer program product. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the porous media CT image segmentation method of this embodiment 1 are loaded.

[0084] A computer-readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, 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.

[0085] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes. 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. 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.

[0086] 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 it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for segmenting porous media CT images, characterized in that, include: Core samples were obtained and scanned using CT equipment to obtain CT images of the core samples. The CT images of the core samples were preprocessed to extract the main part of the gray-scale distribution in the CT images as the images to be analyzed. The image to be analyzed is fitted with a multi-distribution model to obtain a multi-distribution fitted image. Intelligent thresholding segmentation is performed on multi-distribution fitted images to obtain image segmentation results containing the boundaries of different pore filling materials.

2. The porous medium CT image segmentation method according to claim 1, characterized in that, The preprocessing of CT images of core samples to extract the main grayscale distribution from the CT images as the images to be analyzed includes: For a single acquired CT image, a region is cropped at a preset location in the original CT image to obtain the CT image of interest; Gray-scale mapping and main component extraction are performed on the CT images of interest, and the main gray-scale distribution in the extracted CT images is used as the image to be analyzed.

3. The porous medium CT image segmentation method according to claim 1, characterized in that, The process of performing multi-distribution fitting on the image to be analyzed to obtain a multi-distribution fitted image includes: Analyze the grayscale distribution of the image to be analyzed to determine the initial number of components; An improved GMM algorithm is adopted to perform multi-distribution fitting on the image to be analyzed with a number of Cauchy or Gaussian components corresponding to the initial number of components, and to determine the average residual corresponding to the multi-distribution fitting result under the current number of components. Determine whether the average residual corresponding to the multi-distribution fitting result under the current number of components is less than the preset threshold. If so, stop the iteration. Otherwise, update the initial number of components according to the preset step size and re-execute the multi-distribution fitting until the convergence condition is met, and obtain the optimal number of components and its corresponding hyperparameters.

4. The porous medium CT image segmentation method according to claim 3, characterized in that, The improved GMM algorithm is used to perform multi-distribution fitting on the image to be analyzed with a number of Cauchy or Gaussian components corresponding to the initial number of components, and to determine the average residual corresponding to the multi-distribution fitting result under the current number of components, including: ① Determine the range of values ​​for the distribution corresponding to different components based on the characteristics of the rock sample; ② Configure the hyperparameters for the current iteration number based on the defined range of values; ③ Based on the determined hyperparameters, the EM algorithm is used to perform multi-distribution fitting, and the average residual corresponding to the multi-distribution fitting result at the current iteration number is calculated; When using the EM algorithm for multi-distribution fitting, variance constraints are added for each Gaussian or Cauchy component as follows: In the formula, For the first During the nth iteration The variance of the Gaussian component or the first Gaussian component The scale parameter of each Cauchy component This is the lower limit of the variance or scaling parameter. This represents the upper limit of the variance or scale parameter. For the first The grayscale value of a pixel in a Gaussian or Cauchy component; For the first During the nth iteration The weights of Gaussian or Cauchy components; For the first During the nth iteration The mean of the nth Gaussian component or the 1st Gaussian component Position parameters of each Cauchy component; ④ Determine whether the average residual corresponding to the multi-distribution fitting result under the current iteration number is less than the preset threshold. If so, stop the iteration and obtain the optimal hyperparameter corresponding to the current number of components; otherwise, update the hyperparameter according to the preset step size and return to step ③ until the optimal hyperparameter is obtained. ⑤ Take the average residual corresponding to the multi-distribution fitting result of the optimal hyperparameter as the average residual corresponding to the multi-distribution fitting result under the current number of components.

5. The porous medium CT image segmentation method according to claim 1, characterized in that, The intelligent thresholding segmentation of the multi-distribution fitted image to obtain image segmentation results containing boundaries of different pore filling materials includes: Based on Bayesian theory or the likelihood ratio algorithm based on hypothesis testing, all pixels in the multi-distribution fitted image are classified to obtain the preliminary boundaries of different pore filling components. The preliminary boundaries of the different pore-filling components are subjected to morphological examination based on fractal dimension. Based on the examination results, the final binarized image containing the boundaries of different pore-filling components is obtained. The binarized images containing the boundaries of different pore filling materials are optimized to generate high-precision, standardized image segmentation results.

6. The porous medium CT image segmentation method according to claim 5, characterized in that, The classification of all pixels in a multi-distribution fitted image based on Bayesian theory includes: Calculate the probability that each pixel in the multi-distribution fitted image is misclassified as a different pore filling component; Based on the probability value of each pixel being misclassified as a component, all pixels in the multi-distribution fitted image are classified to obtain an image containing the boundaries of different pore filling materials.

7. The porous medium CT image segmentation method according to claim 5, characterized in that, The likelihood ratio algorithm based on hypothesis testing classifies all pixels in a multi-distribution fitted image, including: Calculate the log-likelihood ratio of the probability that each pixel in the multi-distribution fitted image belongs to a specified pore-filling component and other pore-filling components; Based on a given threshold and the calculated log-likelihood ratio, each pixel is assigned to its corresponding pore-filling component, thus completing the classification.

8. A porous media CT image segmentation system, characterized in that, include: The image acquisition module is used to acquire core samples and scan the core samples using CT equipment to obtain CT images of the core samples; The data preprocessing module is used to preprocess the CT image data of the core sample and use the main part of the grayscale distribution in the processed CT image as the image to be analyzed. The multi-distribution fitting module is used to perform multi-distribution fitting on the image to be analyzed, and obtain a multi-distribution fitted image. The intelligent thresholding segmentation unit is used to perform intelligent thresholding segmentation on multi-distribution fitted images to obtain image segmentation results containing the boundaries of different pore filling materials.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the porous media CT image segmentation methods as described in claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the porous media CT image segmentation methods as described in claims 1 to 7.

Images