Coal seam analysis method and system based on image data

Abstract

The application provides a coal seam analysis method and system based on image data, and relates to the technical field of image processing. The method comprises the following steps: acquiring a coal seam image set and performing coal seam structure disassembly on each to-be-analyzed coal seam image to generate a plurality of structure units in the to-be-analyzed coal seam image, determining a local coverage area of each structure unit, constructing group structure features of the plurality of structure units and generating structure matching parameters of each structure unit, performing structure overlap detection on the to-be-analyzed coal seam image, determining a structure abnormal area of the to-be-analyzed coal seam image and performing structure abnormality analysis, generating structure detection information of the structure abnormal area, performing structure change analysis on continuous to-be-analyzed coal seam images according to the structure detection information, and generating a coal seam structure analysis result of the coal seam image set. The application realizes timely and accurate coal seam structure analysis, and meets the real-time and reliability requirements of the coal seam mining process.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for coal seam analysis based on image data. Background Technology

[0002] The identification and analysis of coal seam structures are of great significance in coal mining. With the application of image acquisition equipment in underground operations, coal seam images serve as an important data source for coal seam structure detection and evaluation. Analyzing coal seam images through visual information helps to achieve intuitive identification and real-time monitoring of coal seam structures.

[0003] Coal seams typically contain various structures such as bedding, fractures, and interbedded rock. Due to the complexity of the coal seam structure and the localized damage caused by excavation, the boundaries between different structural units become blurred, easily leading to overlap or confusion. For example, fractures may intersect or be parallel to bedding boundaries, and interbedded rock may be embedded between bedding layers and partially overlap with bedding boundaries, making coal seam structure identification difficult. In actual mining operations, failure to identify and analyze these structures promptly and accurately can easily lead to misjudgments of the coal seam structure, thus affecting mining decisions and operational safety. Some coal seam image analysis methods rely on human experience or image processing techniques based on single features, achieving high accuracy in identifying specific structures. However, in actual mining operations, if the relationships between multiple types of structures are not comprehensively analyzed promptly and accurately, misjudgments of the coal seam structure can easily occur, thus impacting mining decisions. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a coal seam analysis method and system based on image data, which is used to identify different structural information in coal seam images and further detect and analyze areas with interaction or overlap, thereby meeting the real-time and reliability requirements of actual coal seam mining processes.

[0005] The first aspect of this invention provides a coal seam analysis method based on image data, comprising:

[0006] A coal seam image set is acquired, which includes multiple consecutive coal seam images to be analyzed. The coal seam structure is decomposed for each coal seam image to generate multiple structural units in the coal seam image to be analyzed. The structural units include bedding units, fracture units and interbedded rock units.

[0007] Determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features;

[0008] Based on the local coverage area, structural overlap detection is performed on the coal seam image to be analyzed to identify structurally abnormal regions in the coal seam image to be analyzed. Based on the structural matching parameters, structural anomaly analysis is performed on multiple structurally abnormal regions to generate structural detection information of structurally abnormal regions.

[0009] Based on the structural detection information, structural change analysis is performed on continuous coal seam images to generate coal seam structure analysis results for the coal seam image set.

[0010] Preferably, the coal seam structure of each coal seam image to be analyzed is decomposed, and multiple structural units in the coal seam image to be analyzed are extracted, including:

[0011] The coal seam image to be analyzed is subjected to directional filtering to obtain multiple directional energy features. The global bedding direction of the coal seam image to be analyzed is determined based on the directional energy features. The global bedding direction is then used to perform periodic analysis on the coal seam image to be analyzed, generating multiple bedding units of the coal seam image to be analyzed.

[0012] Linear enhancement is performed on the coal seam image to be analyzed to generate a linear response image. Non-maximum suppression processing and connectivity analysis are performed on the linear response image to identify candidate fractures in the coal seam image to be analyzed and generate multiple fracture units.

[0013] Edge detection is performed on the coal seam image to be analyzed to generate an edge feature image. Multiple closed contours are extracted from the edge feature image, and candidate rock-clamping regions are determined in the edge feature image to generate multiple rock-clamping units.

[0014] Preferably, the coal seam image to be analyzed is periodically analyzed based on the global bedding direction to determine multiple bedding units of the coal seam image, including:

[0015] The normal of the global bedding direction is determined and multiple candidate periodic signals of the coal seam image to be analyzed are extracted. Autocorrelation analysis is performed on the multiple candidate periodic signals to determine the local period of the multiple candidate periodic signals. Based on the multiple local periods, the target bedding distribution period of the coal seam image to be analyzed is determined.

[0016] Peak detection is performed on each candidate periodic signal to generate multiple peak nodes. Based on the global bedding direction, the multiple peak nodes are stitched together to generate multiple local bedding sequences of the coal seam image to be analyzed. Based on the target bedding distribution period, fracture detection is performed on each local bedding sequence to generate multiple bedding units of the coal seam image to be analyzed.

[0017] Preferably, structural anomaly regions in the coal seam image to be analyzed are identified, and structural anomaly analysis is performed on multiple structural anomaly regions based on structural matching parameters, including:

[0018] Based on the local coverage area of ​​structural units, multiple structural units are mapped to determine the overlapping area between any two structural units, generating a structural overlap image corresponding to the coal seam image to be analyzed, and extracting multiple structural anomaly areas from the structural overlap image.

[0019] Statistical analysis is performed on each structural anomaly region to construct a pixel distribution matrix for each type of structural unit. Based on the pixel distribution matrix and structural matching parameters, structural coverage parameters and structural distribution parameters for each type of structural unit are generated. Based on the structural coverage parameters and structural distribution parameters, structural confidence parameters for each type of structural unit are calculated. Based on the structural confidence parameters of layered units and interbedded units, the global morphological structure of the structural anomaly region is determined. Based on the structural confidence parameters of crack units, the local morphological structure of the structural anomaly region is determined. The structural detection information of the structural anomaly region is then output.

[0020] Preferably, structural change analysis is performed on continuous coal seam images based on structural detection information, including:

[0021] Based on the structural detection information, a structural distribution image of each coal seam image to be analyzed is generated for each type of structural unit. A structural change set for each type of structural unit is constructed, and the morphological change sequence and positional change sequence of each structural change set are extracted. Based on the morphological change sequence and positional change sequence, the coal seam structural analysis results of the coal seam image set are generated.

[0022] Preferably, a group structural feature of multiple structural units is constructed, and structural matching parameters for each structural unit are generated based on the group structural feature, including:

[0023] For the structural matching parameters, the layer width of each layered element is determined, the average width of multiple layered elements is calculated, and the structural matching parameters of the layered elements are calculated based on the layer width and the average width. The group strength threshold for the interbedded elements is determined, the individual strength parameter of each interbedded element is calculated, and the structural matching parameters of the interbedded elements are calculated based on the individual strength parameters and the strength threshold.

[0024] Preferably, multiple structurally anomalous regions are extracted from the structurally overlapping image, including:

[0025] A sliding window process is applied to the structurally overlapping images to determine the number of overlapping pixels in each window. Window regions with a number of overlapping pixels greater than a preset structural anomaly threshold are marked as structural anomaly regions.

[0026] A second aspect of the present invention provides a coal seam analysis system based on image data, for implementing the aforementioned coal seam analysis method based on image data, comprising:

[0027] The coal seam structure decomposition module is used to acquire a coal seam image set, which includes multiple consecutive coal seam images to be analyzed. The coal seam structure is decomposed for each coal seam image to generate multiple structural units in the coal seam image to be analyzed. The structural units include bedding units, fracture units and interbedded rock units.

[0028] The structural matching analysis module is used to determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features.

[0029] The abnormal region detection module is used to perform structural overlap detection on the coal seam image to be analyzed based on the local coverage area, determine the structural abnormal regions of the coal seam image to be analyzed, perform structural abnormality analysis on multiple structural abnormal regions based on structural matching parameters, and generate structural detection information of structural abnormal regions.

[0030] The structural change analysis module is used to perform structural change analysis on continuous coal seam images based on structural detection information, and generate coal seam structural analysis results for the coal seam image set.

[0031] The present invention has the following beneficial effects:

[0032] This invention acquires and analyzes continuous coal seam images to perform structural decomposition, accurately extracting structural units such as bedding, cracks, and interbedded rock. It combines the local coverage area of ​​structural units with the characteristics of the group structure to generate structural matching parameters for the units. Abnormal regions are located through structural overlap detection. By fusing structural matching parameters with pixel distribution analysis, the global and local morphological confidence of abnormal regions is quantified and analyzed. Based on the structural detection information from multiple frames of images, changes in coal seam structure are dynamically tracked, generating analysis results covering both morphological and positional changes. This allows for timely and accurate comprehensive analysis of the relationships between multiple types of structures, meeting the real-time and reliability requirements of actual coal seam mining processes and providing reliable data support for coal seam stability assessment and intelligent mining decision-making. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating a coal seam analysis method based on image data, provided as an embodiment of the present invention.

[0034] Figure 2 This is a schematic diagram of the structure of a coal seam analysis system based on image data, provided as an embodiment of the present invention. Detailed Implementation

[0035] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention.

[0036] In coal mining operations, images of the coal seam surface can be acquired using face imaging equipment, borehole endoscopes, or vision acquisition devices mounted on robots. Complex environmental factors, such as insufficient lighting, dust interference, and the complex structure of the coal seam, result in inconsistent image quality, and the boundaries of the coal seam structure are prone to blurring, overlapping, or even fracturing. Segmentation methods that rely on manual observation or single features are prone to misjudgment in such scenarios and lack a comprehensive analysis of structural relationships. Therefore, this invention proposes a coal seam analysis method based on image data to achieve unified analysis of multiple structural units in coal seam images, detection and anomaly analysis of overlapping areas, and structural evolution analysis of continuous images during the mining process.

[0037] One embodiment of the present invention provides a coal seam analysis method based on image data; please refer to [link to relevant documentation]. Figure 1 The method includes the following steps:

[0038] Step S1: Obtain a coal seam image set, which includes multiple consecutive coal seam images to be analyzed. Decompose the coal seam structure in each coal seam image to generate multiple structural units in the coal seam image to be analyzed.

[0039] Specifically, information about the coal mining process is collected using imaging devices, and a coal seam image set is constructed based on continuous images acquired during the mining process. Due to the significant layered distribution characteristics of coal seams, strip-like bedding is typically visible in the images, and linear or fracture-like cracks and interbedded rock may also be present. Each coal seam image to be analyzed is structurally decomposed to obtain multiple structural units, including but not limited to bedding units, fracture units, and interbedded rock units. Each structural unit corresponds to an independent structural region in the coal seam image, used to characterize the constituent elements within the coal seam.

[0040] As one implementation process, the process of deconstructing the coal seam structure of each coal seam image to be analyzed and extracting multiple structural units from the coal seam image includes:

[0041] For the extraction of bedding units, considering the obvious layered depositional characteristics of coal seams, which often appear as parallel or approximately parallel stripes in images, directional filtering can be performed on the coal seam image to obtain energy features in multiple directions. Directional filters, such as Gabor filters, can be used to calculate the energy corresponding to the coal seam image at different angles, outputting multiple directional energy features. By comparing the energy distribution in different directions, the direction with the largest energy peak is selected as the global bedding direction of the coal seam image.

[0042] Coal seam bedding is typically represented in images as striped grayscale or texture variations distributed along a global direction. To quantitatively describe the distribution pattern of bedding, periodic analysis is performed on the coal seam image to be analyzed based on the global bedding direction. Specifically, multiple candidate periodic signals are extracted in the direction perpendicular to the global bedding direction. Each candidate periodic signal corresponds to a grayscale sequence of pixel grayscale value changes intercepted along the normal at different positions. Autocorrelation analysis is performed on these candidate periodic signals to detect repetitive patterns and identify the overall spacing characteristics of the bedding structure in the image. Finally, multiple bedding units of the coal seam image to be analyzed are constructed.

[0043] Specifically, after determining the normal of the global bedding direction, a signal point is determined according to a specific positional distribution pattern, such as every several pixels. Candidate periodic signals of each signal point are extracted along the normal of the global bedding direction to obtain multiple candidate periodic signals of the coal seam image to be analyzed. Autocorrelation analysis is then performed on the multiple candidate periodic signals to detect repetitive patterns and determine the local period of each candidate periodic signal. That is, the periodic parameters corresponding to the candidate periodic signals generated after autocorrelation analysis can be statistically analyzed based on multiple local periods to calculate their mean and determine the target bedding distribution period of the coal seam image to be analyzed, which is used to describe the overall spacing characteristics of the bedding structure in the image.

[0044] After extracting the target bedding distribution period, peak detection is performed on each candidate period signal to generate multiple peak nodes, which reflect the boundary positions of bedding strips. Then, bedding regions are stitched together based on the global bedding direction. During the stitching process, multiple initial bedding points are first determined. For example, by analyzing multiple peak nodes of the candidate period signals, the candidate period signal with the most uniform distribution, such as the smallest variance in the distance between peak nodes, is selected as the representative. The multiple peak nodes of the candidate period signals are used as initial bedding points. Combined with the global bedding direction, the peak node of each candidate period signal closest to the straight line formed by the initial bedding point and the global bedding direction is extracted. This constitutes the local bedding sequence corresponding to the initial bedding point. In this way, multiple local bedding sequences of the coal seam image to be analyzed can be initially constructed. Simultaneously, a fracture test is performed on each local bedding sequence in conjunction with the target bedding distribution period. Specifically, the detection process can analyze the distance between two adjacent pixels in the local bedding sequence under the normal angle of the global bedding direction, which is no more than half of the target bedding distribution period. If it exceeds half of the target bedding distribution period, it indicates that the two pixels may not belong to the same bedding zone, indicating that there may be a fracture or anomaly at this point, and the bedding is truncated. Finally, multiple bedding units are generated based on multiple local bedding sequences. Each bedding unit may be a complete local bedding sequence or a segment obtained from a local bedding sequence. In this way, multiple bedding strip structures of the coal seam image to be analyzed are extracted.

[0045] For fracture unit extraction, fractures in coal seams typically exhibit slender, near-linear structural features with significant abrupt changes in grayscale or texture. Therefore, this embodiment performs linear enhancement processing on the coal seam image to strengthen linear feature information. For example, the Hessian matrix of the image can be calculated, and its eigenvalues ​​analyzed. For slender line regions, the Hessian matrix usually shows a difference in eigenvalues ​​between large and small values, which can highlight fracture lines and weaken nonlinear regions, thereby generating a linear response image corresponding to the coal seam image to be analyzed. Subsequently, non-maximum suppression processing is applied to the linear response image to eliminate redundant responses, retaining only the most significant line structures. Then, through connectivity analysis, continuous or near-continuous lines are identified as candidate fractures, thereby generating multiple fracture units and obtaining a fracture structure with continuity and directionality in the coal seam image to be analyzed.

[0046] For the extraction of interbedded rock units, heterogeneous layers such as mudstone and sandstone are often interbedded during coal seam deposition, forming strip-shaped or blocky interbedded rock. In visual images, interbedded rock typically appear as closed regions with significant differences in grayscale and texture from the coal seam background. In this embodiment, edge detection is performed on the coal seam image to be analyzed to generate an edge feature image. Then, multiple closed contours are extracted from the edge feature image as candidate interbedded rock regions, thereby generating multiple interbedded rock units. The interbedded rock regions are segmented by detecting closed contours, avoiding confusion with bedding boundaries or fracture features.

[0047] Step S2: Determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features.

[0048] Specifically, for different structural units, the corresponding local coverage area is determined based on their morphological characteristics. For bedding units, which contain a pixel distribution sequence reflecting the boundary positions of bedding strips, the area sandwiched between two adjacent bedding units can be considered the local coverage area of ​​the bedding unit along the global bedding direction. For interbedded units, the area they cover is directly considered the local coverage area. In coal seam images, the morphology of a single structural unit may be irregular, leading to the introduction of a group structural feature modeling process. By performing a holistic analysis of multiple units within the same image, the overall morphology of the structural unit is determined, generating the group structural features of the structural units. Furthermore, for different individuals, the differences between their individual features and the group structure are analyzed, and the structural matching parameters of the structural unit are calculated. Finally, for different structural units, the corresponding local coverage area is determined based on their morphological characteristics.

[0049] In this embodiment, the extraction of group structure features and the calculation and generation of structure matching parameters are analyzed using layered units and interbedded units, which have more regional coverage significance. For crack units, the focus is on characteristics such as the density of crack distribution.

[0050] For the structural matching parameters of bedding units, coal seam bedding typically exhibits a banded distribution, and its bedding width is a crucial indicator of bedding stability and consistency. For multiple bedding units extracted from the coal seam image to be analyzed, the bedding width of each unit is first calculated. Specifically, this can be achieved by taking the width of multiple pixels of the bedding unit in the normal direction of the global bedding direction within its local coverage area, and calculating the average of these pixels to obtain the bedding width. The average width of multiple bedding units is then used as the group structural feature of the bedding units in the coal seam image. The ratio between the bedding width of each unit and the average width of multiple bedding units is used as the structural matching parameter of the bedding unit. A small difference between the bedding width of a bedding unit and its average width indicates that the bedding unit is consistent with the overall distribution and belongs to a stable unit. Based on this, corresponding structural matching parameters are generated for each bedding unit to characterize the differences between different bedding units and the group structure.

[0051] For the structural matching parameters of interbedded rock units, interbedded rock, as heterogeneous bodies interspersed within coal seams, typically exhibits significantly different grayscale or texture compared to the coal seam, and can therefore be characterized by intensity features. To this end, after extracting multiple interbedded rock units, a group intensity threshold for these units is first determined. This threshold can be obtained through statistical analysis of the grayscale distribution across the entire image, serving as a group structural feature. The group intensity threshold can be understood as the boundary between conventional coal seams and interbedded rock. Coal seams themselves tend to be darker with lower grayscale values; values ​​exceeding the group intensity threshold are considered to be interbedded rock structures belonging to impurities such as mudstone, sandstone, or stony materials. Subsequently, the individual strength parameter of each interbedded rock unit is calculated. In this process, it is considered that if the individual strength parameter is higher than the group strength threshold, it can be regarded as an interbedded rock region. If the individual strength parameter is not less than the group strength threshold, the structural matching parameter of the interbedded rock unit is recorded as 1. Otherwise, the ratio of the individual strength parameter to the group strength threshold is calculated as the structural matching parameter of the interbedded rock unit. When the individual strength parameter is less than the group strength threshold, the greater the deviation from the group strength threshold, the more it indicates that the interbedded rock unit may have abnormality or identification deviation in nature, and the lower the degree of structural matching with the group structural characteristics.

[0052] Step S3: Perform structural overlap detection on the coal seam image to be analyzed based on the local coverage area, determine the structural anomaly region of the coal seam image to be analyzed, perform structural anomaly analysis on multiple structural anomaly regions based on structural matching parameters, and generate structural detection information of structural anomaly regions.

[0053] Specifically, in complex coal seam images, cracks easily intersect with stratification boundaries, and interbedded rock may also partially overlap with stratification boundaries. Failure to distinguish between these can lead to misjudgments of the coal seam structure. This invention, based on locally covered areas, detects overlapping regions between different units to identify multiple structural anomaly regions. Subsequently, by combining structural matching parameters, attribute analysis is performed on the anomaly regions to ultimately generate structural detection information, thereby providing an objective description of structural anomalies present in the coal seam image.

[0054] As one implementation process, structural anomaly regions in the coal seam image to be analyzed are identified, and structural anomaly analysis is performed on multiple structural anomaly regions based on structural matching parameters, including:

[0055] Based on the local coverage area of ​​structural units, multiple structural units are mapped to determine the overlapping area between any two structural units, generating a structural overlap image corresponding to the coal seam image to be analyzed, and extracting multiple structural anomaly areas from the structural overlap image.

[0056] Specifically, multiple structural units are mapped onto the same image coordinate system, forming an overall spatial distribution map. By analyzing the geometric relationships between different structural units, the overlapping areas between any two structural units can be determined. For example, crack units may intersect with the boundary regions of bedding units, and inclusion units may partially overlap with bedding units. After summing these overlapping areas, a structural overlap image corresponding to the original image is generated. A sliding window process is applied to the structural overlap image, and the number of overlapping pixels in the window region is counted. Multiple window regions are marked according to a preset structural anomaly threshold. If the number of overlapping pixels in a window region is greater than the preset structural anomaly threshold, the region corresponding to that window is marked as a structural anomaly region, representing a spatial location where there is a judgment conflict or structural confusion.

[0057] Statistical analysis is performed on each structural anomaly region to construct a pixel distribution matrix for each type of structural unit. Based on the pixel distribution matrix and structural matching parameters, structural coverage parameters and structural distribution parameters for each type of structural unit are generated. Based on the structural coverage parameters and structural distribution parameters, structural confidence parameters for each type of structural unit are calculated.

[0058] Specifically, statistical analysis is performed on each structural region to generate a pixel distribution matrix for each type of structural unit in the structural anomaly region. This matrix reflects the overall coverage ratio and spatial distribution pattern of each type of structural unit in the anomaly region. Further analysis of the pixel distribution matrix yields the structural coverage parameter and structural distribution parameter for each type of structural unit. The structural coverage parameter measures the overall coverage of structural units within the anomaly region and is characterized by the pixel percentage of each structural unit. In the specific calculation process, considering the individual differences among structural units of the same type, the number of pixels belonging to the same structural unit is corrected by the structural matching parameter before being used for counting. For example, if a certain interlocking unit has 10 pixels in the structural anomaly region and the corresponding structural matching parameter is 0.8, the number of pixels is weighted and corrected using the structural matching parameter before being used as the percentage of pixels involved in that interlocking unit for further calculation. The structural distribution parameter measures the spatial distribution state of structural units within the anomaly region and is characterized by the entropy value of the pixel distribution matrix. In calculating the entropy of the pixel distribution matrix, the values ​​of different pixels in the original pixel distribution matrix are represented by 0 and 1 to indicate whether they exist or not; for example, if they exist, they are recorded as 1. In this embodiment, the existence state is corrected by combining the structure matching parameter. The structure matching parameter is used as the existence probability to update the pixel distribution matrix, and then the structure distribution parameter is obtained by calculating the entropy of the pixel distribution matrix. Finally, the product of the structure coverage parameter and the structure distribution parameter is used as the structure confidence parameter of the structural unit.

[0059] The global morphological structure of the structural anomaly region is determined based on the structural confidence parameters of the stratification element and the inclusion element, and the local morphological structure of the structural anomaly region is determined based on the structural confidence parameters of the crack element. The structural detection information of the structural anomaly region is then output.

[0060] Specifically, structural confidence parameters reflect the credibility and dominance of structural units within structurally anomalous regions. For example, if a certain type of unit has a large coverage ratio and a uniform distribution, the structural confidence parameter is high. The morphological structure of the anomalous region is comprehensively analyzed based on the confidence parameters of various structural units. For the global morphological structure, it is mainly determined based on the structural confidence parameters of bedding units and interbedded rock units. If the confidence level of bedding units is high, the anomalous region may primarily belong to bedding structures; if the confidence level of interbedded rock units is high, the region is more likely an extension of interbedded rock bodies. For the local morphological structure, it is mainly determined based on the structural confidence parameters of fracture units. If the confidence level of fracture units in this region is high, it indicates significant fracture interference in the anomalous region. By analyzing the morphology of structurally anomalous regions from both a global structural coverage perspective and a local fracture penetration perspective, the global perspective can determine the main structure of the coal seam region, while the local perspective can determine the density of fractures, assisting in judging the local stability of the coal seam, thereby generating structural detection information for different structurally anomalous regions.

[0061] Step S4: Perform structural change analysis on continuous coal seam images to be analyzed based on structural detection information, and generate coal seam structure analysis results for the coal seam image set.

[0062] Specifically, for coal seam image sets acquired during coal mining as excavation progresses, further analysis is conducted on the continuous spatial changes in the structure of multiple coal seam images. Based on single-frame structural analysis, structural detection information is combined to perform cross-frame correlation of multiple images, tracking the extension, splitting, or disappearance of the same structural unit. For example, the analysis can reveal the spatial variation trend of layer thickness, the propagation path of cracks in multiple images, and the continuity of interbedded rock. The final structural analysis results of the coal seam image set provide reliable data support for coal seam stability evaluation and mining decisions.

[0063] Specifically, based on structural detection information, a structural distribution image for each type of structural unit in each coal seam image to be analyzed can be generated. This structural distribution image represents the overall distribution of a specific type of structural unit. For example, a crack distribution image can characterize the distribution of a single coal seam image, thus constructing a structural change set representing continuous structural changes for each type of structural unit. Spatial change trend analysis can be performed from the perspectives of morphological and positional changes of structural units. For example, for cracks, the difference in the number of cracked windows in two adjacent crack distribution images can be analyzed to characterize whether the crack's affected area is increasing, and whether the center point of the cracked window has shifted, determining whether the structure's position has shifted with excavation recommendations. This extracts the morphological and positional change sequences for each structural change set. For bedding structures, the bedding width change sequence can be further calculated. Finally, based on multiple change sequences, the coal seam structural analysis results for the coal seam image set are generated. For example, crack units may show an expansion trend, bedding units may show a thickness change trend, and interbedded rock units may show an outward expansion trend within a certain area. This provides important data support for safety monitoring and mining progress assessment in coal mining, helping to evaluate the stability and mining feasibility of coal seams in real time.

[0064] This invention provides a coal seam analysis system based on image data, specifically for implementing the aforementioned coal seam analysis method based on image data. Please refer to [link to relevant documentation]. Figure 2 ,include:

[0065] The coal seam structure decomposition module is used to acquire a coal seam image set, which includes multiple consecutive coal seam images to be analyzed. The coal seam structure is decomposed for each coal seam image to generate multiple structural units in the coal seam image to be analyzed. The structural units include bedding units, fracture units and interbedded rock units.

[0066] The structural matching analysis module is used to determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features.

[0067] The abnormal region detection module is used to perform structural overlap detection on the coal seam image to be analyzed based on the local coverage area, determine the structural abnormal regions of the coal seam image to be analyzed, perform structural abnormality analysis on multiple structural abnormal regions based on structural matching parameters, and generate structural detection information of structural abnormal regions.

[0068] The structural change analysis module is used to perform structural change analysis on continuous coal seam images based on structural detection information, and generate coal seam structural analysis results for the coal seam image set.

[0069] The above are merely specific embodiments of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A coal seam analysis method based on image data, characterized in that, include: A coal seam image set is acquired, which includes multiple consecutive coal seam images to be analyzed. The coal seam structure is decomposed for each coal seam image to generate multiple structural units in the coal seam image to be analyzed. The structural units include bedding units, fracture units and interbedded rock units. Determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features; Based on the local coverage area, structural overlap detection is performed on the coal seam image to be analyzed. Multiple structural units are mapped based on the local coverage area of ​​structural units to determine the overlapping area between any two structural units, generating a structural overlap image corresponding to the coal seam image to be analyzed, and extracting multiple structural anomaly areas from the structural overlap image. Statistical analysis is performed on each structural anomaly region to construct a pixel distribution matrix for each type of structural unit. Based on the pixel distribution matrix and structural matching parameters, structural coverage parameters and structural distribution parameters for each type of structural unit are generated. Based on the structural coverage parameters and structural distribution parameters, structural confidence parameters for each type of structural unit are calculated. Based on the structural confidence parameters of layered units and interbedded units, the global morphological structure of the structural anomaly region is determined. Based on the structural confidence parameters of crack units, the local morphological structure of the structural anomaly region is determined. The structural detection information of the structural anomaly region is output. Based on the structural detection information, structural change analysis is performed on continuous coal seam images to generate coal seam structure analysis results for the coal seam image set.

2. The coal seam analysis method based on image data according to claim 1, characterized in that, For each coal seam image to be analyzed, the coal seam structure is decomposed, and multiple structural units are extracted from the coal seam image, including: The coal seam image to be analyzed is subjected to directional filtering to obtain multiple directional energy features. The global bedding direction of the coal seam image to be analyzed is determined based on the directional energy features. The global bedding direction is then used to perform periodic analysis on the coal seam image to be analyzed, generating multiple bedding units of the coal seam image to be analyzed. Linear enhancement is performed on the coal seam image to be analyzed to generate a linear response image. Non-maximum suppression processing and connectivity analysis are performed on the linear response image to identify candidate fractures in the coal seam image to be analyzed and generate multiple fracture units. Edge detection is performed on the coal seam image to be analyzed to generate an edge feature image. Multiple closed contours are extracted from the edge feature image, and candidate rock-clamping regions are determined in the edge feature image to generate multiple rock-clamping units.

3. The coal seam analysis method based on image data according to claim 2, characterized in that, Periodic analysis of the coal seam image under analysis is performed based on the global bedding direction to determine multiple bedding units in the coal seam image, including: The normal of the global bedding direction is determined and multiple candidate periodic signals of the coal seam image to be analyzed are extracted. Autocorrelation analysis is performed on the multiple candidate periodic signals to determine the local period of the multiple candidate periodic signals. Based on the multiple local periods, the target bedding distribution period of the coal seam image to be analyzed is determined. Peak detection is performed on each candidate periodic signal to generate multiple peak nodes. Based on the global bedding direction, the multiple peak nodes are stitched together to generate multiple local bedding sequences of the coal seam image to be analyzed. Based on the target bedding distribution period, fracture detection is performed on each local bedding sequence to generate multiple bedding units of the coal seam image to be analyzed.

4. The coal seam analysis method based on image data according to claim 3, characterized in that, Structural change analysis is performed on continuous coal seam images based on structural detection information, including: Based on the structural detection information, a structural distribution image of each coal seam image to be analyzed is generated for each type of structural unit. A structural change set for each type of structural unit is constructed, and the morphological change sequence and positional change sequence of each structural change set are extracted. Based on the morphological change sequence and positional change sequence, the coal seam structural analysis results of the coal seam image set are generated.

5. The coal seam analysis method based on image data according to claim 3, characterized in that, Construct a group structural feature of multiple structural units, and generate structural matching parameters for each structural unit based on the group structural feature, including: For the structural matching parameters, the layer width of each layered element is determined, the average width of multiple layered elements is calculated, and the structural matching parameters of the layered elements are calculated based on the layer width and the average width. The group strength threshold for the interbedded elements is determined, the individual strength parameter of each interbedded element is calculated, and the structural matching parameters of the interbedded elements are calculated based on the individual strength parameters and the strength threshold.

6. The coal seam analysis method based on image data according to claim 3, characterized in that, Multiple structural anomalous regions were extracted from the structurally overlapping image, including: A sliding window process is applied to the structurally overlapping images to determine the number of overlapping pixels in each window. Window regions with a number of overlapping pixels greater than a preset structural anomaly threshold are marked as structural anomaly regions.

7. A coal seam analysis system based on image data, characterized in that, The system is used to implement the coal seam analysis method based on image data as described in any one of claims 1-6, comprising: The coal seam structure decomposition module is used to acquire a coal seam image set, which includes multiple consecutive coal seam images to be analyzed. The coal seam structure is decomposed for each coal seam image to generate multiple structural units in the coal seam image to be analyzed. The structural units include bedding units, fracture units and interbedded rock units. The structural matching analysis module is used to determine the local coverage area of ​​each structural unit, construct the group structural features of multiple structural units, and generate the structural matching parameters of each structural unit based on the group structural features. The abnormal region detection module is used to perform structural overlap detection on the coal seam image to be analyzed based on the local coverage area, determine the structural abnormal regions of the coal seam image to be analyzed, perform structural abnormality analysis on multiple structural abnormal regions based on structural matching parameters, and generate structural detection information of structural abnormal regions. The structural change analysis module is used to perform structural change analysis on continuous coal seam images based on structural detection information, and generate coal seam structural analysis results for the coal seam image set.

Images