Coal and rock identification method and system based on full-view microscopic images

Abstract

The present disclosure provides a coal rock identification method and system based on full-view microscopic images, which relates to the technical field of coal rock identification, including: obtaining coal rock microscopic images; extracting overlapping areas of adjacent coal rock microscopic images after preprocessing, using a full-view feature point matching image stitching method to match feature points in the overlapping areas, and finding matching point pairs in the overlapping areas; using a geometric change estimation method to transform and align matching point pairs of different images, using a weighted average image fusion method to stitch images, and generate a full-view image; inputting the full-view image into a component identification model, extracting multi-scale semantic features of the full-view image, and reconstructing the spatial information of the multi-scale semantic features, and outputting the identified coal rock microscopic component category results; quantitatively analyzing the identified coal rock microscopic components, obtaining the proportion of each component and summarizing the statistics. The present disclosure can intelligently identify the type of coal rock based on multi-scale features.

Description

Technical Field

[0001] The present disclosure relates to the technical field of coal rock identification, and in particular to a coal rock identification method and system based on full-viewing-field microscopic images. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Coal lithology identification is crucial for assessing coal quality, predicting its combustion characteristics, and determining its economic value. This technology helps understand how the characteristics of different coal types affect their calorific value, combustion efficiency, and ash production, thereby optimizing coal utilization and processing, reducing environmental pollution, and providing a scientific basis for coal mining and resource management. Traditional coal lithology identification relies on microscopic observation and analysis, with manual determination of coal types such as lignite, bituminous coal, and anthracite. While this method can provide detailed identification information, the identification process is highly dependent on the operator's experience and judgment and is susceptible to subjective factors. Furthermore, manual identification is slow and labor-intensive when processing large numbers of samples, and standards and results can be inconsistent between different operators.

[0004] With the continuous development of artificial intelligence technology, computer vision has shown broad application prospects in the field of coal and rock microscopic image recognition. It can automatically process and analyze large quantities of microscopic images, significantly improving recognition speed and efficiency while reducing human error. Leveraging advanced algorithms such as deep learning, recognition accuracy and reliability have been significantly improved.

[0005] In recent years, image segmentation technology has been widely used in coal and rock microscopic image recognition, achieving significant progress. Semantic segmentation methods, by classifying each pixel in microscopic images, can accurately distinguish between different coal and rock components, thereby enabling automated identification of coal and rock types, significantly improving recognition accuracy and efficiency. This technology not only significantly reduces the workload of traditional manual labeling but also effectively avoids human error, enabling automated processing of large-scale samples.

[0006] However, the current lack of standardized datasets for coal and rock microscopic images limits the algorithm's versatility and reliability. Furthermore, deep learning models capable of quantitatively analyzing coal and rock microscopic images have yet to be developed. Furthermore, intelligent coal and rock identification systems require further refinement. Summary of the Invention

[0007] In order to solve the above problems, the present disclosure proposes a coal rock identification method and system based on full-view microscopic images, which adopts full-view feature point matching image splicing to match the overlapping areas of adjacent coal rock microscopic images, and splice them to form a full-view digital image. Based on the image segmentation algorithm, the spliced ​​full-view image is automatically segmented to achieve accurate identification of the microscopic components of the coal rock, and quantitative analysis of the identified microscopic components of the coal rock, thereby realizing automatic identification of the coal rock types and facilitating subsequent data retrieval, management and application.

[0008] According to some embodiments, the present disclosure adopts the following technical solutions:

[0009] The coal and rock identification method based on full-view microscopic images includes:

[0010] Acquire coal and rock microscopic images and pre-process the coal and rock microscopic images;

[0011] The overlapping areas of adjacent coal and rock microscopic images after preprocessing are extracted, and the full-view feature point matching image stitching method is used to match the feature points of the overlapping areas and find the matching point pairs in the overlapping areas.

[0012] The matching point pairs of different images are transformed and aligned using the geometric change estimation method, and the image fusion method of weighted average is used to stitch the images to generate a full-view image.

[0013] The full-view image is input into the component recognition model to extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the classification results of the identified coal and rock microscopic components;

[0014] Conduct quantitative analysis on the identified coal rock micro-components, obtain the proportion of each component and summarize the statistics;

[0015] The type of coal rock can be determined based on the distribution and proportion of microscopic components.

[0016] According to some embodiments, the present disclosure adopts the following technical solutions:

[0017] The coal and rock identification system based on full-view microscopic images includes:

[0018] An image preprocessing module is used to obtain and preprocess the coal and rock microscopic images;

[0019] The image stitching module is used to extract the overlapping areas of adjacent coal and rock microscopic images after preprocessing, match the feature points of the overlapping areas using the full-view feature point matching image stitching method, and find the matching point pairs in the overlapping areas; use the geometric change estimation method to transform and align the matching point pairs of different images, and use the weighted average image fusion method to stitch the images to generate a full-view image;

[0020] The microscopic component recognition module is used to input the full-view image into the component recognition model, extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the recognized coal and rock microscopic component classification results;

[0021] The quantitative analysis module is used to quantitatively analyze the identified microscopic components of coal and rock, obtain the proportion of each component and summarize the statistics;

[0022] The coal rock intelligent identification module is used to determine the type of coal rock based on the distribution and proportion of each component.

[0023] According to some embodiments, the present disclosure adopts the following technical solutions:

[0024] A non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the coal and rock identification method based on full-viewing-field microscopic images is implemented.

[0025] According to some embodiments, the present disclosure adopts the following technical solutions:

[0026] An electronic device comprises: a processor, a memory and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device implements the coal and rock identification method based on full-field microscopic images.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The disclosed coal and rock identification method based on full-view microscopic images utilizes automated image preprocessing methods to rapidly perform noise removal, contrast enhancement, and other processes, significantly improving image quality and reducing the time required for manual intervention. Furthermore, the ability to quickly and accurately process large numbers of images significantly enhances recognition efficiency and speed.

[0029] The disclosed coal and rock identification method based on full-view microscopic images uses a full-view feature point matching image stitching method to match feature points in overlapping areas, identify matching point pairs within the overlapping areas, transform and align matching point pairs from different images using a geometric change estimation method, and stitch them together using a weighted average image fusion method. This method integrates microscopic images from multiple fields of view into a single large image, enabling analysis on a wider scale. Through a deep learning model, it is possible to extract and analyze microscopic components at multiple scales, identifying complex coal and rock structures.

[0030] The coal rock identification method based on full-field microscopic images disclosed in the present invention can accurately calculate the proportion of each type of component through an automated quantitative analysis method, realize quantitative analysis, and thus realize automatic identification of coal rock types, reduce errors in manual calculations and reduce labor intensity.

[0031] The disclosed coal and rock identification method based on full-view microscopic images integrates data storage and management methods, enabling effective management and long-term preservation of analytical data, facilitating subsequent query and research. Furthermore, the system possesses excellent scalability, adapting to diverse coal and rock intelligent identification needs through model updates or additional training data. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings, which constitute a part of the present disclosure, are used to provide a further understanding of the present disclosure. The exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation to the present disclosure.

[0033] Figure 1 This is a flow chart of a method for identifying coal and rock based on full-view microscopic images according to an embodiment of the present disclosure;

[0034] Figure 2 This is a diagram of a coal and rock identification system based on full-viewing field microscopic images according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0035] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0036] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present disclosure belongs.

[0037] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0038] Example 1

[0039] In one embodiment of the present disclosure, a method for identifying coal and rock based on a full-viewing-field microscopic image is provided, comprising: step 1: acquiring a coal and rock microscopic image, and preprocessing the coal and rock microscopic image;

[0040] Step 2: Extract the overlapping areas of adjacent coal and rock microscopic images after preprocessing, use the full-view feature point matching image stitching method to match the feature points of the overlapping areas, and find the matching point pairs in the overlapping areas;

[0041] Step 3: Use the geometric change estimation method to transform and align the matching point pairs of different images, and use the weighted average image fusion method to perform image stitching to generate a full-view image;

[0042] Step 4: Input the full-view image into the component recognition model, extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the recognized coal and rock microscopic component classification results;

[0043] Step 5: Quantitatively analyze the identified coal rock microscopic components, obtain the proportion of each component and summarize the statistics.

[0044] Step 6: Determine the type of coal rock based on the distribution and proportion of microscopic components.

[0045] Among them, lignite has a low vitrinite content and contains more inertinite and minerals; bituminous coal has a higher vitrinite content and a compact structure; anthracite has the highest vitrinite content, a high reflectivity, and very little inertinite.

[0046] If the vitrinite content is >80%, the reflectivity is high and the inertinite is very small, it can be identified as anthracite; if the vitrinite content is 40-80%, the inertinite is moderate, it is bituminous coal; if the vitrinite content is <40%, there are more minerals and inertinite, it is lignite.

[0047] As an embodiment, the specific implementation process of the coal rock identification method based on full-view microscopic images disclosed in the present invention is as follows:

[0048] Step 1: Obtaining a coal rock microscopic image and preprocessing the coal rock microscopic image;

[0049] Specifically, a coal rock microscopic image is obtained and preprocessed, including image denoising and contrast enhancement processing.

[0050] Step 2: Extract the overlapping areas of adjacent coal and rock microscopic images after preprocessing, use the full-view feature point matching image stitching method to match the feature points of the overlapping areas, and find the matching point pairs in the overlapping areas;

[0051] Specifically, it includes: 1) Feature point detection: The SIFT feature point detection algorithm is used to find unique and repeatable feature points in each image. This method can extract feature points and descriptors in the image under scale and rotation invariance for subsequent matching process:

[0052]

[0053] Among them, by adjusting the standard deviation To change the degree of blur, through the multiple relationship k To generate images of different scales, usually k is set to ,in s is the number of image layers in each scale group, is the Gaussian difference function at different scales, is the Gaussian blur function, is the input image.

[0054] 2) Feature point matching: Use Euclidean distance to calculate the distance between the descriptors of each feature point in adjacent images, and select the feature point with the smallest distance as the matching pair. This method can pair the feature points in adjacent images and find matching point pairs in the overlapping area:

[0055]

[0056] in, It is the descriptor of two feature points.

[0057] Step 3: Use the geometric change estimation method to transform and align the matching point pairs of different images, and use the weighted average image fusion method to perform image stitching to generate a full-view image;

[0058] Specifically, 1) Geometric transformation estimation: Use homography transformation to determine the geometric transformation relationship between adjacent coal and rock microscopic images, and obtain the transformation matrix so that the feature points in one coal and rock microscopic image are aligned with the feature points in another coal and rock microscopic image through transformation:

[0059]

[0060] in, and are the coordinates of the points in the original image and the transformed image respectively, H for The homography matrix of .

[0061] 2) Image stitching: Apply the estimated transformation matrix to transform and align the images, and then use the weighted average image fusion method to reduce stitching gaps and color differences:

[0062]

[0063] in, is the fused image, , For the images involved in stitching, , is the weighting coefficient.

[0064] Repeat steps 2 and 3 for all adjacent coal and rock microscopic images, gradually stitching the images together to generate a full-viewing-area image.

[0065] Step 4: Input the full-view image into the component recognition model, extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the recognized coal rock microscopic component classification results;

[0066] Specifically, the component recognition model adopts the DeepLabV3+ network based on the encoder-decoder architecture, and uses the component recognition model to automatically segment the spliced ​​full-view image to achieve accurate identification of the microscopic components of coal and rock.

[0067] First, the multi-scale semantic features of the image are extracted in the encoder, and then the spatial information of these features is reconstructed in the decoder, so that the predicted image output by the network is consistent with the original input image in spatial resolution.

[0068] The input data for the component recognition model is a full-view image of the microscopic components of coal and rock. In the encoder, the backbone network is responsible for preliminary feature extraction. The backbone network uses the first five convolutional modules of ResNet-101. The first module is a 7x7 convolutional layer, and the remaining four modules are composed of stacked residual bottleneck structures with varying numbers of residuals. Residual learning effectively prevents the gradient vanishing and gradient exploding phenomena that occur with increasing network layers, thereby helping the network to more comprehensively extract local detailed features in the image.

[0069] However, relying solely on local detail features is insufficient to fully identify microscopic components. To this end, a dilated spatial pyramid pooling (ASPP) module is introduced after the backbone network to extract more statistically significant deep semantic information.

[0070] ASPP replaces traditional convolution with dilated convolution, which can significantly expand the receptive field without increasing model parameters and complexity, thereby reducing the loss of spatial resolution. This means that dilated convolution can efficiently extract dense features on images of any resolution. On this basis, ASPP uses four parallel dilated convolution layers, including a 1x1 convolution layer and three 3x3 convolution layers, with expansion rates set to 6, 12, and 18 respectively, and combines global average pooling and convolution fusion to obtain large-scale, multi-scale semantic features. This multi-scale feature contains global and local information of objects in the image, which can help the model simultaneously capture the information brought by the size differences of different microscopic components, helping the model better understand the content of the image.

[0071] However, the feature information extracted by ASPP is relatively abstract and has a large number of channels, lacking local details such as image edge features. Therefore, before these features are input to the decoder, they are first passed through a 1x1 convolution layer to reduce the number of channels. They are then fused with the low-level image features extracted by the backbone network to supplement the missing local details. For the fused feature map, a 3x3 convolution is used to further reduce the dimension of the feature channels. Then, bilinear interpolation upsampling is used to restore the feature map to the same size as the original input image, generating the microscopic component segmentation result. This segmentation result is actually a set of multi-channel microscopic component category probability maps, where each channel corresponds to a recognizable microscopic component category and each pixel value in the channel represents the probability that the pixel belongs to a certain category.

[0072] Furthermore, the microscopic components in the coal rock are identified to include vitrinite, inertinite, exinite, background resin, etc.

[0073] Step 5: Quantitatively analyze the identified coal rock microscopic components, obtain the proportion of each component and summarize the statistics.

[0074] Specifically, the identified coal rock microscopic components are quantitatively analyzed to provide the proportion of each component and other relevant data to support further research and analysis. The segmented image is statistically classified and the total number of pixels in each category is calculated. N categories, for each category i , respectively count the number of pixels belonging to this category in the segmented image .

[0075] Calculate the total number of pixels in the entire image ;

[0076] The proportion of each component Calculated by the following formula:

[0077]

[0078] in, Indicates the i The proportion of class components, represents the number of pixels of the i-th component, Indicates the total number of pixels in the image.

[0079] The proportion of each type of microscopic component is summarized as the final statistical result, and the type of coal is inferred based on the distribution and proportion of each component: if the vitrinite content is >80%, the reflectivity is high and the inertinite is very small, it can be identified as anthracite; if the vitrinite content is 40-80% and the inertinite is moderate, it is bituminous coal; if the vitrinite content is <40% and there are more minerals and inertinite, it is lignite.

[0080] Finally, the recognition results are automatically saved in the coal rock microscopic image database to facilitate subsequent data retrieval, management and application.

[0081] Example 2

[0082] In one embodiment of the present disclosure, a coal and rock identification system based on full-view microscopic images is provided, comprising:

[0083] An image preprocessing module is used to obtain and preprocess the coal and rock microscopic images;

[0084] The image stitching module is used to extract the overlapping areas of adjacent coal and rock microscopic images after preprocessing, match the feature points of the overlapping areas using the full-view feature point matching image stitching method, and find the matching point pairs in the overlapping areas; use the geometric change estimation method to transform and align the matching point pairs of different images, and use the weighted average image fusion method to stitch the images to generate a full-view image;

[0085] The microscopic component recognition module is used to input the full-view image into the component recognition model, extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the recognized coal and rock microscopic component classification results;

[0086] The quantitative analysis module is used to quantitatively analyze the identified coal rock microscopic components, obtain the proportion of each component and summarize the statistics.

[0087] The intelligent coal rock identification module is used to infer the type of coal based on the distribution and proportion of each component: if the vitrinite content is >80%, the reflectivity is high and the inertinite is very small, it can be identified as anthracite; if the vitrinite content is 40-80%, and the inertinite is moderate, it is bituminous coal; if the vitrinite content is <40%, and there are more minerals and inertinite, it is lignite.

[0088] Furthermore, it also includes a data storage and management module for automatically saving the identification and analysis results into the coal rock microscopic component image database to facilitate subsequent data retrieval, management and application.

[0089] As an embodiment, a method for using a coal rock identification system based on full-view microscopic images includes:

[0090] A. Obtaining microscopic images of coal rocks.

[0091] B. Use the image preprocessing module to preprocess the coal rock microscopic image, including denoising and contrast enhancement processing.

[0092] C. Use the image stitching module to stitch the pre-processed coal and rock microscopic images to generate a full-viewing-field image.

[0093] D. Use the microscopic component identification module and the deep learning model to identify the microscopic components in coal rocks, such as vitrinite, inertinite, exinite, background resin, etc.

[0094] E. Use the quantitative analysis module to quantitatively analyze the identified coal rock microscopic components and provide the proportion of each component and other relevant data.

[0095] F. Determine the type of coal rock based on the distribution and proportion of microscopic components.

[0096] G. Automatically save the recognition results to the coal rock microscopic image dataset to facilitate subsequent data retrieval, management and application.

[0097] H. Repeat steps AG until all coal rock microcomponents are identified.

[0098] Example 3

[0099] In one embodiment of the present disclosure, a non-transitory computer-readable storage medium is provided, which is used to store computer instructions. When the computer instructions are executed by a processor, the coal and rock identification method based on full-viewing-field microscopic images is implemented.

[0100] Example 4

[0101] In one embodiment of the present disclosure, an electronic device is provided, comprising: a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the coal rock identification method based on full-field microscopic images.

[0102] The present disclosure is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0103] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0104] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Those skilled in the art should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A coal-rock identification method based on full-view microscopic images, characterized in that: include: Acquire coal and rock microscopic images and pre-process the coal and rock microscopic images; The overlapping areas of adjacent coal and rock microscopic images after preprocessing are extracted, and the full-view feature point matching image stitching method is used to match the feature points of the overlapping areas and find the matching point pairs in the overlapping areas. The matching point pairs of different images are transformed and aligned using the geometric change estimation method, and the image fusion method of weighted average is used to stitch the images to generate a full-view image. The full-view image is input into the component recognition model to extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the classification results of the identified coal and rock microscopic components; Conduct quantitative analysis on the identified coal rock microscopic components, obtain the proportion of each component and summarize the statistics; Determine the type of coal rock based on the distribution and proportion of microscopic components; Among them, the component recognition model adopts the DeepLabV3+ network based on the encoder-decoder architecture, and uses the component recognition model to automatically segment the spliced ​​full-view image. First, the multi-scale semantic features of the image are extracted in the encoder, and then the spatial information of the multi-scale semantic features is reconstructed in the decoder. Finally, the predicted image output by the network is consistent with the original input image in spatial resolution. In the encoder, the backbone network is responsible for preliminary feature extraction. The backbone network selects the first five convolution modules of ResNet-101, of which the first module is a 7x7 convolution layer, and the remaining four modules are stacked by different numbers of residual bottleneck structures. The hollow spatial pyramid pooling ASPP module is introduced after the backbone network to extract more statistically significant deep-level semantic information. ASPP uses four parallel hollow convolution layers, including a 1x1 convolution layer and three 3x3 convolution layers, with expansion rates of respectively. Set to 6, 12 and 18, and combined with global average pooling and convolution fusion to obtain large-scale, multi-scale semantic features. This multi-scale feature contains global and local information of objects in the image, which can help the model simultaneously capture the information brought by the size differences of different microscopic components. Before these multi-scale semantic features are input into the decoder, they are first required to pass through a 1x1 convolution layer to reduce the number of channels, and then fused with the low-level image features extracted by the backbone network to supplement the missing local detail information. For the fused feature map, 3x3 convolution is used to further reduce the dimension of the feature channel, and then bilinear interpolation upsampling is used to restore the feature map to the same size as the original input image to generate the segmentation result of the microscopic component. The segmentation result is a set of multi-channel microscopic component category attribution probability maps. Each channel corresponds to a identifiable microscopic component category, and each pixel value on the channel represents the probability that the pixel belongs to a certain category.

2. The coal-rock identification method based on full-view microscopic images according to claim 1, characterized in that: After extracting the overlapping areas of adjacent coal rock microscopic images after preprocessing, the SIFT feature point detection algorithm is first used to find unique and repeatable feature points in each image, and then the feature points and descriptors in the coal rock microscopic component images are extracted under scale and rotation invariance: in, is the Gaussian difference function at different scales, is the Gaussian blur function, is the input image, is the standard deviation, It is a multiple relationship.

3. The coal-rock identification method based on full-view microscopic images according to claim 1, characterized in that: The full-view feature point matching image stitching method is used to match feature points in the overlapping area, including: using Euclidean distance to calculate the distance between the descriptors of each feature point in adjacent coal and rock microscopic images, selecting the feature point with the smallest distance as the matching pair, and thus pairing the feature points in adjacent coal and rock microscopic images to find the matching point pairs in the overlapping area.

4. The coal-rock identification method based on full-view microscopic images according to claim 1, characterized in that: The matching point pairs of different images are transformed and aligned using the geometric change estimation method, including: using homography transformation to determine the geometric transformation relationship between adjacent coal and rock microscopic images, obtaining the transformation matrix, so that the feature points in one coal and rock microscopic image are aligned with the feature points in another coal and rock microscopic image through transformation.

5. The coal-rock identification method based on full-view microscopic images according to claim 1, characterized in that: Use the weighted average image fusion method to stitch images and generate a full-view image, including: transforming and aligning the images using the estimated transformation matrix, and using the weighted average image fusion method to reduce stitching gaps and color differences: in, is the fused image, , For the images involved in stitching, , is the weighting coefficient.

6. The method for identifying coal and rock based on full-view microscopic images according to claim 1, characterized in that: Quantitative analysis of the identified coal rock microscopic components includes: pixel classification statistics of the output microscopic component category attribution probability map, calculation of the total number of pixels in each category, summarizing the proportion of each category of microscopic components as the final statistical result, and judging the type of coal rock based on the distribution and proportion of the microscopic components.

7. A coal-rock identification system based on full-view microscopic images, used to implement the coal-rock identification method based on full-view microscopic images according to any one of claims 1 to 6, characterized in that: include: An image preprocessing module is used to obtain and preprocess the coal and rock microscopic images; The image stitching module is used to extract the overlapping areas of adjacent coal and rock microscopic images after preprocessing, match the feature points of the overlapping areas using the full-view feature point matching image stitching method, and find the matching point pairs in the overlapping areas; use the geometric change estimation method to transform and align the matching point pairs of different images, and use the weighted average image fusion method to stitch the images to generate a full-view image; The microscopic component recognition module is used to input the full-view image into the component recognition model, extract the multi-scale semantic features of the full-view image, reconstruct the spatial information of the multi-scale semantic features, and output the recognized coal and rock microscopic component classification results; The quantitative analysis module is used to quantitatively analyze the identified coal rock microscopic components, obtain the proportion of each component and summarize the statistics; The coal rock intelligent identification module is used to infer the type of coal based on the distribution and proportion of each component.

8. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the coal and rock identification method based on full-viewing-field microscopic images as described in any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the coal rock identification method based on full-field microscopic images as described in any one of claims 1 to 6.

Images