Shale electron microscope scanning processing method and equipment on logging site and storage medium thereof

Abstract

The invention provides a shale electron microscope scanning processing method and device for a logging site and a storage medium, and the method comprises the steps: carrying out the comprehensive scanning of a target shale in a low-resolution form, and obtaining a first image of the target shale; based on an image optimization model generated by training of an adversarial neural network architecture, performing high-resolution conversion on the first image to obtain a second image; determining a high-value area of the target shale in the first image; scanning the high-value region in a high-resolution form to obtain a third image of the target shale; determining an overlapping position of the third image in the second image based on the high-value region; and according to the overlapping position, the second image and the third image are spliced and verified to obtain the target image of the target shale, and high-resolution conversion is carried out on the image by using an image optimization model through a combination mode of low-resolution whole sample scanning and high-resolution local scanning, so that the scanning efficiency is improved, and meanwhile, the precision of a global image is ensured.

Description

Technical Field

[0001] The present application relates to the technical field of electron microscopes, and in particular to a shale electron microscope scanning and processing method, equipment and storage medium thereof at a logging site. Background Art

[0002] In the development of unconventional oil and gas resources such as shale gas, it is crucial to accurately obtain the microscopic pore structure and organic matter distribution of rocks. Scanning electron microscopes (SEMs) are widely used for microstructural analysis of rock samples due to their high resolution. However, existing SEM scanning methods usually use fixed resolution scanning line by line and pixel by pixel, which is extremely inefficient and time-consuming when processing large-area samples. Especially in the scanning of complex rock structures such as shale samples, it is difficult to balance image accuracy and scanning efficiency. Summary of the invention

[0003] The main purpose of the embodiments of the present application is to propose a shale electron microscope scanning and processing method, equipment and storage medium for logging site, which combines low-resolution full-sample scanning with high-resolution local scanning, and uses an image optimization model with an adversarial neural network architecture to perform high-resolution conversion of images, thereby greatly reducing the scanning time of shale, improving the scanning efficiency, and ensuring the accuracy and details of the global image.

[0004] To achieve the above-mentioned purpose, a first aspect of an embodiment of the present application proposes a shale electron microscope scanning and processing method at a logging site, comprising: Performing a comprehensive scan of the target shale in a low-resolution form to obtain a first image of the target shale, where the first image is a low-resolution overall real image; Based on the image optimization model trained and generated by the adversarial neural network architecture, the first image is converted to a high-resolution image to obtain a second image, where the second image is a high-resolution overall virtual image; identifying high-value areas of target shale in the first image; Scanning high-value areas in high-resolution form to obtain a third image of the target shale, the third image being a high-resolution local true image; Based on the high value area, determining an overlapping position of the third image in the second image; According to the overlapping position, the second image and the third image are spliced ​​and verified to obtain a target image of the target shale, which is a high-resolution overall image.

[0005] Furthermore, in some embodiments, the adversarial neural network architecture of the image optimization model includes a generator including a multi-layer neural network and a discriminator including a multi-layer neural network, and the training generation process of the image optimization model includes: Acquire a first sample image of a plurality of shale samples, where the first sample image is a high-resolution overall real image; Performing low-resolution processing on the plurality of first sample images to obtain a plurality of second sample images; Initialize the adversarial neural network architecture and build the original image model based on the adversarial neural network architecture; According to the multiple first sample images and the multiple second sample images, adversarial iterative training is performed on the original image model to obtain an image optimization model.

[0006] Further, in some embodiments, in the process of performing low-resolution processing on a plurality of first sample images to obtain a plurality of second sample images, performing low-resolution processing on each first sample image includes: Based on the down sampling method, pixel merging is performed on the first sample image to obtain a first low-resolution sample image; Performing Gaussian blurring and adding noise to the first low-resolution sample image to obtain a second low-resolution sample image; Performing sub-image segmentation on the second low-resolution sample image to obtain a plurality of third low-resolution sample images, wherein the third low-resolution sample images are sub-images of the second low-resolution sample image; Converting the pixel domain of each third low-resolution sample image into a corresponding frequency domain to obtain a plurality of frequency domain image data; By using a lossy quantization matrix, high-frequency data in each frequency domain image data is filtered to obtain a plurality of low-frequency image data; Based on the compression coding algorithm, multiple low-frequency image data are combined and compressed to obtain a second sample image.

[0007] Further, in some embodiments, in the process of performing multiple adversarial iterative training on the image original model according to the multiple first sample images and the multiple second sample images to obtain the image optimization model, each time the adversarial iterative training of the image original model is performed, the adversarial iterative training includes: Inputting the plurality of second sample images into the generator to increase the image resolution, thereby obtaining a plurality of third sample images, wherein the third sample images are high-resolution overall virtual images; Inputting a plurality of third sample images and a plurality of first sample images into a discriminator to obtain a gradient loss value of a current iteration, wherein the gradient loss value is used to feedback the difference between the third sample image and the first sample image; According to the gradient loss value, the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator are back-propagated and updated based on the Adam optimization algorithm; If the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator in the current iterative training meet the preset training requirements, the iterative training of the original image model is terminated to obtain the image optimization model.

[0008] Further, in some embodiments, the plurality of second sample images are input into the generator to increase the image resolution, thereby obtaining a plurality of third sample images, including: Performing feature extraction on the plurality of second sample images through the convolution layer of the generator to obtain a plurality of first feature images; Performing high-resolution simulation on the multiple first feature images through an upsampling layer of the generator to obtain multiple high-resolution sample images; In the batch normalization layer of the generator, multiple high-resolution sample images are nonlinearly processed through an activation function to obtain multiple third sample images.

[0009] Further, in some embodiments, the plurality of third sample images and the plurality of first sample images are input into the discriminator to obtain the gradient loss value of the current iteration, including: Through the convolution layer of the discriminator, feature extraction is performed on the plurality of first sample images to obtain a plurality of second feature images; Performing feature extraction on the plurality of third sample images through the convolution layer of the discriminator to obtain a plurality of third feature images; In the fully connected layer of the discriminator, the difference between the multiple second feature images and the multiple third feature images is calculated by the mean square error loss function to obtain the gradient loss value.

[0010] Further, in some embodiments, the second image and the third image are spliced ​​and verified according to the overlapping position to obtain a target image of the target shale, including: According to the feature point detection algorithm, feature points at overlapping positions are detected in the second image to obtain a first feature point set; Detecting feature points of the third image according to a feature point detection algorithm to obtain a second feature point set; Determine similar feature points between the first feature point set and the second feature point set by a feature point matching algorithm to obtain a plurality of feature point matching pairs, wherein the feature point matching pairs are used to associate similar feature points between the second image and the third image; By using an image transformation algorithm, a spatial transformation relationship between a plurality of feature point matching pairs is calculated to obtain a transformation relationship matrix between the second image and the third image; In the second image, aligning the third image with the overlapping position; Based on the transformation relationship matrix, the third image is mapped to the overlapping position to obtain the target image.

[0011] Further, in some embodiments, determining a high-value area of ​​target shale in the first image includes: Extracting feature points in the first image according to a feature point detection algorithm to obtain a fourth feature point set; Performing spatial clustering on the fourth feature point set by a density clustering algorithm to obtain a plurality of feature point clusters in the first image; The regions of multiple feature point clusters are marked as high-value regions.

[0012] To achieve the above-mentioned purpose, the second aspect of an embodiment of the present application proposes an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the shale electron microscope scanning processing method of the above-mentioned first aspect embodiment when executing the computer program.

[0013] To achieve the above-mentioned purpose, the third aspect of the embodiment of the present application proposes a storage medium, which is a computer-readable storage medium. The storage medium stores a computer program, and when the computer program is executed by the processor, the shale electron microscope scanning processing method of the above-mentioned first aspect embodiment is implemented.

[0014] The embodiments of the present application have the following beneficial effects: the present application obtains a first image of the target shale by comprehensively scanning the target shale in a low-resolution form, and the first image is a low-resolution overall real image; based on the image optimization model trained and generated by the adversarial neural network architecture, the first image is converted to high resolution to obtain a second image, and the second image is a high-resolution overall virtual image; the high-value area of ​​the target shale is determined in the first image; the high-value area is scanned in a high-resolution form to obtain a third image of the target shale, and the third image is a high-resolution local real image; based on the high-value area, the overlapping position of the third image in the second image is determined; according to the overlapping position, the second image and the third image are spliced ​​and verified to obtain a target image of the target shale, and the target image is a high-resolution overall real image, and then through the combination of low-resolution full-sample scanning and high-resolution local scanning, and the image is converted to high resolution using the image optimization model of the adversarial neural network architecture, the scanning time of the shale is greatly reduced, the scanning efficiency is improved, and the accuracy and details of the global image are ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of a shale electron microscope scanning and processing method at a logging site provided by some embodiments of the present application; Figure 2 is a flowchart of a training image optimization model provided by some embodiments of the present application; Figure 3 is a flow chart of low-resolution processing of a first sample image provided by some embodiments of the present application; Figure 4 is a flowchart of performing adversarial iterative training of an image original model each time provided by some embodiments of the present application; Figure 5 Some embodiments of the present application provide Figure 4 Flow chart of step S401 in FIG. Figure 6 Some embodiments of the present application provide Figure 4 Flow chart of step S402; Figure 7 Some embodiments of the present application provide Figure 1 Flow chart of step S106; Figure 8 Some embodiments of the present application provide Figure 1 Flow chart of step S103; Fig. 9 A schematic diagram of the structure of a shale electron microscope scanning optimization system provided in some embodiments of the present application; Fig.10 It is a schematic diagram of the hardware structure of an electronic device provided in some embodiments of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0017] In the description of the present application, it should be understood that descriptions involving orientation, such as up, down, front, back, left, right, etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0018] It should also be noted that in the description of this application, "several" means more than one, "more" means more than two, "greater than", "less than", "exceed", etc. are understood to exclude the number itself, and "above", "below", "within", etc. are understood to include the number itself. If there is a description of "first" or "second", it is only used to distinguish the technical features, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0020] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0021] In the development of unconventional oil and gas resources such as shale gas, it is crucial to accurately obtain the microscopic pore structure and organic matter distribution of rocks. Scanning electron microscopes (SEMs) are widely used for microstructural analysis of rock samples due to their high resolution. However, existing SEM scanning methods usually use fixed resolution scanning line by line and pixel by pixel, which is extremely inefficient and time-consuming when processing large-area samples. Especially in the scanning of complex rock structures such as shale samples, it is difficult to balance image accuracy and scanning efficiency.

[0022] Based on this, the embodiments of the present application provide a shale electron microscope scanning and processing method, equipment and storage medium for logging site, which combines low-resolution full-sample scanning with high-resolution local scanning, and uses an image optimization model with an adversarial neural network architecture to perform high-resolution conversion of images, thereby greatly reducing the scanning time of shale and improving the scanning efficiency, while ensuring the accuracy and details of the global image.

[0023] The embodiments of the present application provide a shale electron microscope scanning and processing method, equipment and storage medium at a logging site, which are specifically described through the following embodiments.

[0024] In the first aspect, a shale electron microscope scanning and processing method at a logging site in an embodiment of the present application is first described.

[0025] Reference Figure 1 As shown, Figure 1 It is a flow chart of a shale electron microscope scanning and processing method at a logging site provided by some embodiments of the present application. The shale electron microscope scanning and processing method may include but is not limited to steps S101 to S106.

[0026] Step S101: comprehensively scan the target shale in a low-resolution form to obtain a first image of the target shale.

[0027] The first image is a low-resolution overall real image.

[0028] In a possible implementation, the scanning range of the electron microscope is set, and the target shale is scanned using a low magnification (such as 500 times) to generate a low-resolution first image. The physical size of the first image is consistent with an image obtained by an existing high-resolution SEM scan.

[0029] Step S102: Based on the image optimization model trained and generated by the adversarial neural network architecture, the first image is converted to high resolution to obtain a second image.

[0030] The second image is a high-resolution overall virtual image.

[0031] In one possible implementation, the first image is input into an image optimization model generated by training based on an adversarial neural network architecture to generate a preliminary high-resolution second image.

[0032] Step S103: Determine the high-value area of ​​the target shale in the first image.

[0033] Specifically, feature points are analyzed in a low-resolution first image to capture texture, edge and other structural information in the first image, and then high-value areas of target shale, i.e., key areas of target shale, are determined in the first image based on the texture, edge and other structural information in the first image.

[0034] It should be noted that the number of high-value areas can be one or more, and this application does not make any specific limitation.

[0035] Step S104: Scan the high-value area in a high-resolution form to obtain a third image of the target shale.

[0036] The third image is a high-resolution local real image.

[0037] In a possible implementation, the electron microscope is controlled to scan the high-value region at a high magnification (eg, 50,000 times) to generate a high-resolution third image, the physical size of which is consistent with the size of the high-value region.

[0038] Step S105: Based on the high-value area, determine the overlapping position of the third image in the second image.

[0039] In a possible implementation, since the second image is converted from the first image, the physical size of the second image is consistent with the physical size of the first image, and the overlapping position of the third image in the second image can be determined based on the high-value area of ​​the target shale.

[0040] Step S106: based on the overlapping positions, the second image and the third image are spliced ​​and verified to obtain a target image of the target shale.

[0041] The target image is a high-resolution overall image.

[0042] In a possible implementation, the third image is aligned with the overlapping position in the second image, and then the second image and the third image are spliced ​​and verified to obtain a target image of the target shale. Then, the image can be obtained by local high-resolution scanning, and the high-resolution overall virtual image generated by the image optimization model is further corrected to ensure that the details of the target image are highly consistent with the real image.

[0043] It should be noted that, through step S101 to step S106, the present application can significantly reduce the time of electron microscope scanning and improve the scanning efficiency. Under the same field of view, the scanning time of the present application is only less than 5% of the existing electron microscope scanning method, which saves a lot of scanning time and provides the possibility for high-intensity engineering applications. The scanning process of the present application not only reduces the operating cost, but also improves the overall work efficiency, which is an effective supplement to the existing electron microscope scanning technology.

[0044] Furthermore, the adversarial neural network architecture of the image optimization model includes a generator containing a multi-layer neural network and a discriminator containing a multi-layer neural network, referring to Figure 2 As shown, Figure 2 It is a flowchart of a training image optimization model provided in some embodiments of the present application. The method of training an image optimization model may include but is not limited to steps S201 to S204.

[0045] Step S201: Acquire a first sample image of a plurality of shale samples.

[0046] The first sample image is a high-resolution overall real image. In one possible implementation, a group of 10 shale samples are obtained with high resolution and a wide field of view as a whole (i.e., first sample images). Each first sample image is obtained by stitching together a conventional electron microscope after scanning a sample 2500 times in a row, so that the first sample image has both details and a wide field of view.

[0047] Step S202: performing low-resolution processing on the plurality of first sample images to obtain a plurality of second sample images.

[0048] In a possible implementation, a plurality of first sample images are processed at a low resolution to obtain a plurality of second sample images, thereby providing a data basis for subsequent training of the original image model.

[0049] Step S203: Initialize the adversarial neural network architecture, and construct an original image model based on the adversarial neural network architecture.

[0050] In one possible implementation, a generator and a discriminator of an adversarial neural network architecture are initialized, wherein the generator is used to generate realistic images based on an input image, and the multi-layer neural network of the generator includes multiple convolutional layers and upsampling layers, while the discriminator distinguishes between generated images and real images, and the multi-layer neural network of the discriminator includes multiple convolutional layers and fully connected layers, and then an original model of the image is constructed based on the adversarial neural network architecture.

[0051] Step S204: According to the multiple first sample images and the multiple second sample images, the original image model is subjected to adversarial iterative training to obtain an image optimization model.

[0052] In a possible implementation, a plurality of first sample images and a plurality of second sample images are input into an image primitive model so that the image primitive model undergoes adversarial iterative training. When the image primitive model training is completed, an image optimization model is obtained.

[0053] It should be noted that, through steps S201 to S204, the image optimization model of the present application adopts an adversarial network architecture and designs a special generator and discriminator, so that low-resolution images can be effectively converted into high-resolution images. This not only optimizes the training process of the image optimization model, but also improves the realism of the generated image through the feedback mechanism of the discriminator, and has high model adaptability and accuracy.

[0054] Reference Figure 3 As shown, Figure 3 This is a flowchart of low-resolution processing of a first sample image provided by some embodiments of the present application. The method of low-resolution processing of the first sample image may also include but is not limited to steps S301 to S306.

[0055] Step S301: Based on a down-sampling method, pixel merging is performed on a first sample image to obtain a first low-resolution sample image.

[0056] In one possible implementation, the downsampling ratio is first determined, that is, the ratio of the size of the low-resolution image to the original image. Common ratios include 2x, 4x, etc., depending on the required resolution. For example, if the original image size is 1024×1024, selecting 2x downsampling will reduce the image to 512×512, and then, the various downsampling areas in the first sample image are divided according to the downsampling ratio, and the pixel values ​​in all pixel blocks in each downsampling area are averaged to obtain the pixel values ​​after the combination of each downsampling area, and the various downsampling areas are combined to obtain the first low-resolution sample image.

[0057] Step S302: Gaussian blur and noise addition are performed on the first low-resolution sample image to obtain a second low-resolution sample image.

[0058] In a possible implementation, first, each pixel of the first low-resolution sample image is passed through a Gaussian blur filter, so that the pixel values ​​of the first low-resolution sample image are weighted averaged, the detail information is reduced, but the overall structure is still maintained to achieve the effect of smoothing the image. This process will make the edges in the image blurred; then, Gaussian noise is added to the first low-resolution sample image, so that the value of each pixel in the first low-resolution sample image is subject to a certain degree of random disturbance, and the image becomes rougher and irregular. Among them, Gaussian noise is a random noise that conforms to the normal distribution and can simulate many natural and artificial interference sources, thereby making the second low-resolution sample image more natural and realistic.

[0059] Step S303: performing sub-image segmentation on the second low-resolution sample image to obtain a plurality of third low-resolution sample images.

[0060] The third low-resolution sample image is a sub-image of the second low-resolution sample image.

[0061] In a possible implementation, sub-image segmentation is performed on the second low-resolution sample image to obtain a plurality of third low-resolution sample images.

[0062] Step S304: converting the pixel domain of each third low-resolution sample image into a corresponding frequency domain to obtain a plurality of frequency domain image data.

[0063] In a possible implementation, for each third low-resolution sample image, pixel domain data of the third low-resolution sample image is converted into frequency domain data by two-dimensional Fourier transform to obtain a plurality of frequency domain image data.

[0064] The frequency domain image data obtained after Fourier transform is a complex matrix, in which each element represents the amplitude and phase of the corresponding frequency component.

[0065] Step S305: filtering high-frequency data in each frequency domain image data by using a lossy quantization matrix to obtain a plurality of low-frequency image data.

[0066] Among them, the quantization matrix is ​​a matrix used to quantize frequency domain data. Its element values ​​determine the quantization accuracy of different frequency components. In the lossy quantization matrix, the quantization accuracy of its high-frequency components is usually low (that is, the quantization step is larger), while the quantization accuracy of the low-frequency components is high (that is, the quantization step is smaller).

[0067] In a possible implementation, a matrix operation is performed on the lossy quantization matrix and the frequency domain image data to filter high-frequency data in each frequency domain image data to obtain a plurality of low-frequency image data.

[0068] Step S306: Based on a compression coding algorithm, multiple low-frequency image data are combined and compressed to obtain a second sample image.

[0069] In a possible implementation, based on a compression coding algorithm (such as a JPEG compression coding algorithm, a PNG compression coding algorithm, etc.), image merging and compression are performed on a plurality of low-frequency image data to obtain a second sample image.

[0070] It should be noted that, through steps S301 to S306, the present application combines multiple low-resolution image processing to achieve high-precision correspondence with real high-resolution images, thereby improving the training effect of the model and the quality of the generated images, and filling the deficiencies of the prior art in low-resolution image processing.

[0071] Further, in some embodiments, referring to Figure 4 As shown, Figure 4 This is a flowchart of performing adversarial iterative training on an image original model each time provided by some embodiments of the present application. The method of performing adversarial iterative training on an image original model each time may also include but is not limited to steps S401 to S404.

[0072] Step S401: input a plurality of second sample images into a generator to increase image resolution, thereby obtaining a plurality of third sample images.

[0073] Among them, the third sample image is a high-resolution overall virtual image.

[0074] In one possible implementation, multiple second sample images are input into the generator. After receiving the input, the generator passes through a series of convolutional layers, upsampling layers, or other types of neural network layers to gradually increase the resolution of the image. During this process, the generator attempts to generate an output image similar to a real high-resolution image to obtain multiple third sample images.

[0075] Step S402: Input the plurality of third sample images and the plurality of first sample images into the discriminator to obtain the gradient loss value of the current iteration.

[0076] The gradient loss value is used to feedback the difference between the third sample image and the first sample image.

[0077] In a possible implementation, multiple first sample images are used as positive samples, and multiple third sample images are used as negative samples to be input into the discriminator, and then the discriminator performs a series of convolution, pooling and other operations on the input images, and finally outputs a probability value representing the authenticity of the image. For the real image (i.e., the first sample image), the discriminator should output a probability value close to 1; and for the generated image (i.e., the third sample image), it should output a probability value close to 0. Finally, the probability value of the first sample image and the probability value of the third sample image are cross-calculated through the binary cross entropy loss function to obtain the gradient loss value of the current iteration.

[0078] Step S403: According to the gradient loss value, the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator are back-propagated and updated based on the Adam optimization algorithm.

[0079] In one possible implementation, the initial learning rate (α) of the Adam optimizer, the exponential decay rate of the first-order moment estimate (β1, usually close to 1), the exponential decay rate of the second-order moment estimate (β2, usually close to 1 but less than β1), and a small constant ε (usually set to 1e-8) for numerical stability are initialized. Then, according to the back-propagation algorithm and the gradient loss value, the gradient value of each parameter of the multi-layer neural network in the generator and the discriminator is calculated. Then, according to the current gradient values ​​of each parameter and the previous moment estimate, the estimates of the first-order moment and the second-order moment are updated, and the first-order moment and the second-order moment are biased and corrected. Finally, the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator are updated by back-propagation through the corrected first-order moment and second-order moment, as well as the learning rate.

[0080] Step S404: If the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator in the current iterative training meet the preset training requirements, the iterative training of the image optimization model is terminated to obtain the image optimization model.

[0081] In a possible implementation, if the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator in the current iterative training are both within a certain numerical range in terms of the loss function value, the iterative training of the original image model is terminated to obtain an image optimization model.

[0082] In a possible implementation, if the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator in the current iterative training are both within a certain numerical range in terms of recall rate, the iterative training of the original image model is terminated to obtain an image optimization model.

[0083] Reference Figure 5 As shown, Figure 5 Some embodiments of the present application provide Figure 4In the flowchart of step S401, the method for converting the target chart component into a configuration drawing may include but is not limited to steps S501 to S503.

[0084] Step S501: extract features from a plurality of second sample images through a convolutional layer of a generator to obtain a plurality of first feature images.

[0085] In one possible implementation, multiple second sample images are input into the convolution layer of the generator, and then the convolution layer of the generator performs a convolution operation on the input second sample images, and then each convolution kernel in the convolution layer outputs a feature map (i.e., the first feature image), and these feature maps together constitute the feature representation of the input image.

[0086] Step S502: Perform high-resolution simulation on multiple first feature images through the upsampling layer of the generator to obtain multiple high-resolution sample images.

[0087] In one possible implementation, multiple first feature images are input into the upsampling layer of the generator, the resolution of the first feature images is gradually increased, and more detailed and realistic image content is generated, thereby obtaining multiple high-resolution sample images. This process can be regarded as gradually reconstructing low-resolution feature images into high-resolution images.

[0088] Step S503: In the batch normalization layer of the generator, a plurality of high-resolution sample images are nonlinearly processed by an activation function to obtain a plurality of third sample images.

[0089] In one possible implementation, when a high-resolution sample image is input to a batch normalization layer of a generator, the batch normalization layer uses an activation function (such as a ReLU activation function) to nonlinearly transform the high-resolution sample image, thereby stabilizing the training process of the original image model and improving the quality of the generated image.

[0090] Reference Figure 6 As shown, Figure 6 Some embodiments of the present application provide Figure 4 In the flowchart of step S402, the method for obtaining the target data set may include but is not limited to steps S601 to S603.

[0091] Step S601: extracting features from a plurality of first sample images through the convolutional layer of the discriminator to obtain a plurality of second feature images.

[0092] In a possible implementation, multiple first sample images are input into the convolution layer of the discriminator, and then the convolution layer of the discriminator performs a convolution operation on the input second sample image, and finally each convolution kernel in the convolution layer outputs a feature map (i.e., the second feature image).

[0093] Step S602: extracting features from the plurality of third sample images through the convolutional layer of the discriminator to obtain a plurality of third feature images.

[0094] In one possible implementation, multiple third sample images are input into the convolution layer of the discriminator, and then the convolution layer of the discriminator performs a convolution operation on the input second sample image, and finally each convolution kernel in the convolution layer outputs a feature map (i.e., the third feature image).

[0095] Step S603: In the fully connected layer of the discriminator, the difference between the plurality of second feature images and the plurality of third feature images is calculated by a mean square error loss function to obtain a gradient loss value.

[0096] In one possible implementation, when multiple second feature images and multiple third feature images are input into the fully connected layer of the discriminator, the pixel-level difference between the second feature images and the third feature images is calculated by the mean square error loss function, and finally, the gradient of the mean square error loss value relative to the discriminator parameters is calculated by the back propagation algorithm to obtain the gradient loss value of the current iteration.

[0097] Reference Figure 7 As shown, Figure 7 Some embodiments of the present application provide Figure 1 In the flowchart of step S106, the shale electron microscope scanning processing method may include but is not limited to steps S701 to S706.

[0098] Step S701: Detect feature points at overlapping positions in the second image according to a feature point detection algorithm to obtain a first feature point set.

[0099] In one possible implementation, the Harris corner detection algorithm is used to calculate the degree of change in the horizontal and vertical directions of the grayscale value of each pixel in the overlapping position of the second image, and then determine whether the pixel is a corner point based on these degrees of change. Finally, the corner point is confirmed as a subset of the first feature point set.

[0100] Step S702: Detect feature points of the third image according to a feature point detection algorithm to obtain a second feature point set.

[0101] In one possible implementation, the Harris corner detection algorithm is used to calculate the degree of change in the horizontal and vertical directions of the grayscale value of each pixel in the third image, and then whether the pixel is a corner point is determined based on these degrees of change. Finally, the corner point is confirmed as a subset of the second feature point set.

[0102] Step S703: Determine similar feature points between the first feature point set and the second feature point set through a feature point matching algorithm to obtain a plurality of feature point matching pairs.

[0103] The feature point matching pair is used to associate feature points having similarities between the second image and the third image.

[0104] Specifically, the FLANN feature point matching algorithm determines similar feature points between the first feature point set and the second feature point set to obtain a plurality of feature point matching pairs.

[0105] Step S704: Calculate the spatial transformation relationship between multiple feature point matching pairs through an image transformation algorithm to obtain a transformation relationship matrix between the second image and the third image.

[0106] Specifically, the spatial transformation relationship (including rigid transformation, affine transformation, projection transformation, etc.) between multiple feature point matching pairs is calculated through an image transformation algorithm to obtain a transformation relationship matrix between the second image and the third image.

[0107] Step S705: aligning the third image with the overlapping position in the second image.

[0108] Specifically, the third image is moved above the overlapping position in the second image and aligned.

[0109] Step S706: Based on the transformation relationship matrix, the third image is mapped to the overlapping position to obtain a target image.

[0110] Specifically, based on the transformation relationship matrix, the third image is mapped to the overlapping position, and at the same time, the overlapping position is edge stitched to obtain the target image.

[0111] Reference Figure 8 As shown, Figure 8 Some embodiments of the present application provide Figure 1 In the flowchart of step S103, the shale electron microscope scanning processing method may include but is not limited to steps S801 to S803.

[0112] Step S801: extracting feature points from the first image according to a feature point detection algorithm to obtain a fourth feature point set.

[0113] In one possible implementation, the Harris corner detection algorithm is used to calculate the degree of change in the horizontal and vertical directions of the grayscale value of each pixel in the first image, and then determine whether the pixel is a corner point based on these degrees of change. Finally, the corner point is confirmed as a subset of the fourth feature point set.

[0114] Step S802: spatially clustering the fourth feature point set using a density clustering algorithm to obtain a plurality of feature point clusters in the first image.

[0115] In a possible implementation, the fourth feature point set is spatially clustered using a DBSCAN density clustering algorithm to obtain a plurality of feature point clusters in the first image.

[0116] Step S803: Marking the regions of the plurality of feature point clusters as high-value regions respectively.

[0117] In a possible implementation, regions of multiple feature point clusters are marked as high-value regions respectively.

[0118] In a second aspect, the present application also provides a configuration interface dynamic configuration system, referring to Fig. 9 , Fig. 9 A schematic diagram of the structure of a shale electron microscope scanning optimization system provided in some embodiments of the present application, the shale electron microscope scanning optimization system 900 includes: The image scanning module 901 is configured to perform a comprehensive scan of the target shale in a low-resolution form to obtain a first image of the target shale, where the first image is a low-resolution overall real image; The image generation module 902 is further configured to perform high-resolution conversion on the first image based on the image optimization model trained and generated by the adversarial neural network architecture to obtain a second image, where the second image is a high-resolution overall virtual image; The image generation module 902 is also configured to determine high-value areas of the target shale in the first image; The image scanning module 901 is further configured to scan the high-value area in a high-resolution form to obtain a third image of the target shale, where the third image is a high-resolution local real image; The image generation module 902 is further configured to determine, based on the high value area, an overlapping position of the third image in the second image; The image generation module 902 is further configured to perform splicing verification on the second image and the third image according to the overlapping position to obtain a target image of the target shale, where the target image is a high-resolution overall real image.

[0119] The above-mentioned shale electron microscope scanning optimization system 900 and the above-mentioned shale electron microscope scanning processing method are based on the same inventive concept. The above-mentioned process describes the shale electron microscope scanning optimization system 900 of the embodiment of the present application, which obtains a first image of the target shale by comprehensively scanning the target shale in a low-resolution form, and the first image is a low-resolution overall real image; based on the image optimization model trained and generated by the adversarial neural network architecture, the first image is converted to high resolution to obtain a second image, and the second image is a high-resolution overall virtual image; the high-value area of ​​the target shale is determined in the first image; the high-value area is scanned in a high-resolution form to obtain a third image of the target shale, and the third image is a high-resolution local real image; based on the high-value area, the overlapping position of the third image in the second image is determined; according to the overlapping position, the second image and the third image are spliced ​​and verified to obtain a target image of the target shale, and the target image is a high-resolution overall real image, and then through the combination of low-resolution full sample scanning and high-resolution local scanning, and the image optimization model of the adversarial neural network architecture is used to convert the image to high resolution, thereby greatly reducing the scanning time of the shale, improving the scanning efficiency, and ensuring the accuracy and details of the global image.

[0120] The embodiment of the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned shale electron microscope scanning processing method when executing the computer program. The electronic device can be any smart terminal including a mobile phone, a tablet computer, a car computer, etc.

[0121] See also Fig.10 , Fig.10 : is a schematic diagram of the hardware structure of an electronic device provided in some embodiments of the present application, and the electronic device includes: The processor 1001 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the shale electron microscope scanning processing method provided in the embodiment of the present application; The memory 1002 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 1002 can store an operating system and other application programs. When the technical solution provided in the embodiments of this specification is implemented by software or firmware, the relevant program code is stored in the memory 1002, and the processor 1001 calls and executes the shale electron microscope scanning processing method provided in the embodiments of this application; Input / output interface 1003, used to implement information input and output; Communication interface 1004, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WIFI, Bluetooth, etc.); A bus 1005 , which transmits information between various components of the device (e.g., the processor 1001 , the memory 1002 , the input / output interface 1003 , and the communication interface 1004 ); The processor 1001 , the memory 1002 , the input / output interface 1003 and the communication interface 1004 are connected to each other in communication within the device via the bus 1005 .

[0122] The embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the shale electron microscope scanning processing method provided in the embodiment of the present application is implemented.

[0123] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0124] The embodiments described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application and do not constitute a limitation on the technical solutions provided in the embodiments of the present application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0125] Those skilled in the art will appreciate that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0126] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0127] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0128] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0129] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0130] In the several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0131] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0133] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0134] The preferred embodiments of the present application are described above with reference to the accompanying drawings, but the scope of the rights of the present application is not limited thereto. Any modification, equivalent substitution and improvement made by a person skilled in the art without departing from the scope and essence of the present application should be within the scope of the rights of the present application.

Claims

1. A shale electron microscope scanning and processing method at a logging site, characterized in that: include: Performing a comprehensive scan of the target shale in a low-resolution form to obtain a first image of the target shale, wherein the first image is a low-resolution overall real image; Based on the image optimization model generated by training the adversarial neural network architecture, the first image is converted to a high-resolution image to obtain a second image, where the second image is a high-resolution overall virtual image; determining a high-value area of ​​the target shale in the first image; Scanning the high-value area in a high-resolution form to obtain a third image of the target shale, wherein the third image is a high-resolution local real image; Based on the high-value area, determining an overlapping position of the third image in the second image; According to the overlapping position, the second image and the third image are spliced ​​and verified to obtain a target image of the target shale, and the target image is a high-resolution overall image.

2. The shale electron microscope scanning processing method according to claim 1 is characterized in that: The adversarial neural network architecture of the image optimization model includes a generator including a multi-layer neural network and a discriminator including a multi-layer neural network. The training generation process of the image optimization model includes: Acquire a first sample image of a plurality of shale samples, wherein the first sample image is a high-resolution overall real image; Performing low-resolution processing on the plurality of first sample images to obtain a plurality of second sample images; Initialize the adversarial neural network architecture and build the original image model based on the adversarial neural network architecture; According to the plurality of the first sample images and the plurality of the second sample images, adversarial iterative training is performed on the image original model to obtain the image optimization model.

3. The shale electron microscope scanning processing method according to claim 2 is characterized in that: In the process of performing low-resolution processing on the plurality of first sample images to obtain the plurality of second sample images, performing low-resolution processing on each of the first sample images includes: Based on a down-sampling method, pixel merging is performed on the first sample image to obtain a first low-resolution sample image; Performing Gaussian blur and noise addition on the first low-resolution sample image to obtain a second low-resolution sample image; Performing sub-image segmentation on the second low-resolution sample image to obtain a plurality of third low-resolution sample images, wherein the third low-resolution sample images are sub-images of the second low-resolution sample image; Converting the pixel domain of each of the third low-resolution sample images into a corresponding frequency domain to obtain a plurality of frequency domain image data; Filtering high-frequency data in each of the frequency domain image data by using a lossy quantization matrix to obtain a plurality of low-frequency image data; Based on a compression coding algorithm, multiple low-frequency image data are combined and compressed to obtain the second sample image.

4. The shale electron microscope scanning processing method according to claim 2 is characterized in that: In the process of performing multiple adversarial iterative training on the image original model according to the multiple first sample images and the multiple second sample images to obtain the image optimization model, each time the adversarial iterative training is performed on the image original model, the method includes: Inputting a plurality of the second sample images into the generator to increase the image resolution, thereby obtaining a plurality of third sample images, wherein the third sample images are high-resolution overall virtual images; Inputting a plurality of the third sample images and a plurality of the first sample images into the discriminator to obtain a gradient loss value of a current iteration, wherein the gradient loss value is used to feedback the difference between the third sample image and the first sample image; According to the gradient loss value, back-propagation update is performed on the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator based on the Adam optimization algorithm; If the multi-layer neural network parameters of the generator and the multi-layer neural network parameters of the discriminator in the current iterative training meet the preset training requirements, the iterative training of the original image model is terminated to obtain the image optimization model.

5. The shale electron microscope scanning processing method according to claim 4 is characterized in that: The step of inputting the plurality of second sample images into the generator to increase the image resolution to obtain a plurality of third sample images comprises: Performing feature extraction on the plurality of second sample images through the convolution layer of the generator to obtain a plurality of first feature images; Performing high-resolution simulation on the plurality of first feature images through the upsampling layer of the generator to obtain a plurality of high-resolution sample images; In the batch normalization layer of the generator, the plurality of high-resolution sample images are nonlinearly processed by an activation function to obtain the plurality of third sample images.

6. The shale electron microscope scanning processing method according to claim 4 is characterized in that: The step of inputting the plurality of third sample images and the plurality of first sample images into the discriminator to obtain a gradient loss value of a current iteration includes: Performing feature extraction on the plurality of first sample images through the convolution layer of the discriminator to obtain a plurality of second feature images; Performing feature extraction on the plurality of third sample images through the convolution layer of the discriminator to obtain a plurality of third feature images; In the fully connected layer of the discriminator, the difference between the plurality of second feature images and the plurality of third feature images is calculated by a mean square error loss function to obtain the gradient loss value.

7. The shale electron microscope scanning processing method according to claim 1 is characterized in that: The step of performing splicing verification on the second image and the third image according to the overlapping position to obtain a target image of the target shale includes: According to a feature point detection algorithm, detecting feature points at the overlapping positions in the second image to obtain a first feature point set; Detecting feature points of the third image according to a feature point detection algorithm to obtain a second feature point set; Determine similar feature points between the first feature point set and the second feature point set by a feature point matching algorithm to obtain a plurality of feature point matching pairs, wherein the feature point matching pairs are used to associate similar feature points between the second image and the third image; By using an image transformation algorithm, calculating the spatial transformation relationship between the plurality of feature point matching pairs, and obtaining a transformation relationship matrix between the second image and the third image; In the second image, aligning the third image with the overlapping position; Based on the transformation relationship matrix, the third image is mapped to the overlapping position to obtain the target image.

8. The shale electron microscope scanning processing method according to claim 1 is characterized in that: Determining the high-value area of ​​the target shale in the first image includes: Extracting feature points from the first image according to a feature point detection algorithm to obtain a fourth feature point set; Performing spatial clustering on the fourth feature point set by a density clustering algorithm to obtain a plurality of feature point clusters in the first image; The regions of the plurality of feature point clusters are respectively marked as the high-value regions.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the shale electron microscope scanning processing method according to any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the shale electron microscope scanning processing method according to any one of claims 1 to 8 is implemented.

Images