Underground coal mine anomaly detection method

Abstract

The embodiment of the invention provides an underground coal mine anomaly detection method, and the method comprises the steps: obtaining a Monte dust image collected by an image collection device disposed in an underground coal mine, and determining a Monte dust reference image and a non-Monte dust reference image related to the Monte dust image; determining a Monte dust global pixel value of the Monte dust reference image and a non-Monte dust global pixel value of the non-Monte dust reference image, and calculating a Monte dust elimination feature according to the Monte dust global pixel value and the non-Monte dust global pixel value; updating the Mongolian dust image by using the Mongolian dust elimination feature, and obtaining a target image according to an updating result; and inputting the target image into an anomaly detection model for processing to obtain an anomaly detection result of the coal mine underground area corresponding to the image acquisition equipment.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of image processing technology, and in particular to an anomaly detection method in coal mines. Background Art

[0002] With the development of computer and internet technologies, the traditional coal industry has begun a digital transformation. By integrating and applying key technologies such as the Internet of Things (IoT), cloud computing, big data, and artificial intelligence (AI), breakthroughs have been made in core technologies such as IoT access, intelligent data analysis, three-dimensional visualization, and coordinated management and control of production, transportation, and sales. This has led to the construction of an intelligent, integrated coal mine management and control platform. This has enabled the traditional coal industry to move away from its original management and control model and enter a digital management and control model, effectively promoting the intelligent development of the coal industry and the construction of smart coal mines. To improve the safety of underground coal mine operations, existing cameras are deployed at multiple locations within the mine area for monitoring. When abnormal objects or dangerous accidents occur, alarm strategies can be triggered based on the monitored images or videos, providing timely feedback to operations and maintenance personnel to prevent excessive losses. However, underground coal mine operations can generate dust, which can cause unclear images or videos captured by the cameras. This problem seriously affects anomaly detection results, and an effective solution is urgently needed to address this issue. Summary of the Invention

[0003] In view of this, embodiments of this specification provide a method for detecting anomalies in coal mines. One or more embodiments of this specification also relate to an apparatus for detecting anomalies in coal mines, a computing device, a computer-readable storage medium, and a computer program product to address technical deficiencies in the prior art.

[0004] According to a first aspect of an embodiment of this specification, a method for detecting anomalies in a coal mine is provided, comprising: Acquire a dusty image captured by an image capture device arranged underground in a coal mine, and determine a dusty reference image and a non-dusty reference image associated with the dusty image; determining a dust global pixel value of the dust reference image and a non-dust global pixel value of the non-dust reference image, and calculating a dust removal feature based on the dust global pixel value and the non-dust global pixel value; updating the dust image using the dust removal feature, and obtaining a target image according to the update result; The target image is input into an anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground coal mine area.

[0005] According to a second aspect of the embodiments of this specification, there is provided an anomaly detection device in a coal mine, comprising: an acquisition module configured to acquire a dusty image acquired by an image acquisition device arranged underground in a coal mine, and determine a dusty reference image and a non-dusty reference image associated with the dusty image; a determination module configured to determine a dust global pixel value of the dust reference image and a non-dust global pixel value of the non-dust reference image, and calculate a dust removal feature based on the dust global pixel value and the non-dust global pixel value; an updating module configured to update the dust image using the dust removal feature and obtain a target image according to an updating result; The processing module is configured to input the target image into the anomaly detection model for processing, and obtain the anomaly detection result of the image acquisition device corresponding to the underground coal mine area.

[0006] According to a third aspect of the embodiments of this specification, a computing device is provided, including: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above-mentioned anomaly detection method in coal mines are implemented.

[0007] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned anomaly detection method in a coal mine are implemented.

[0008] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, comprising a computer program or instructions, which, when executed by a processor, implement the steps of the above-mentioned anomaly detection method in a coal mine.

[0009] The anomaly detection method for underground coal mines provided by the present embodiment can, in order to eliminate the dust problem existing in the images captured by the image acquisition device, first determine the dust reference image and the non-dust reference image associated with the dust image after obtaining the dust image captured by the image acquisition device arranged underground in the coal mine, so that the dust reference image and the non-dust reference image can be used to characterize the difference before and after dusting. Therefore, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined, and the dust elimination feature is calculated on this basis. The dust elimination feature can eliminate the influence of dust on the dust image, so that the dust elimination feature can be used to update the dust image, and the target image is obtained according to the update result; the target image is the image with the dust elimination effect, that is, the target image is closer to the actual situation in the real coal mine. Thereafter, the target image can be input into the anomaly detection model for processing to obtain the anomaly detection result of the image acquisition device corresponding to the underground coal mine area. After the image acquisition device captures the image but before performing anomaly detection, the collected image can be dust-removed, which can make the image after the dust effect is eliminated closer to the real situation, thereby effectively improving the accuracy of subsequent anomaly detection in coal mines, making underground coal mine operations safer and more reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a flow chart of a method for detecting anomalies in a coal mine provided by one embodiment of this specification; Figure 2 This is a schematic diagram of an image in a method for detecting anomalies in a coal mine provided by one embodiment of this specification; Figure 3 This is a flowchart of a processing process of an anomaly detection method in a coal mine provided by an embodiment of this specification; Figure 4 This is a schematic structural diagram of an anomaly detection device for an underground coal mine provided by one embodiment of this specification; Figure 5 This is a structural block diagram of a computing device provided by one embodiment of this specification. DETAILED DESCRIPTION

[0011] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0012] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "the," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0013] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0014] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0015] This specification provides a method for detecting anomalies in a coal mine. One or more embodiments of this specification also relate to an apparatus for detecting anomalies in a coal mine, a computing device, a computer-readable storage medium, and a computer program product, each of which is described in detail in the following embodiments.

[0016] See also Figure 1 , Figure 1 A flow chart of an anomaly detection method in a coal mine provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0017] Step S102: obtaining a dusty image captured by an image capture device arranged underground in a coal mine, and determining a dusty reference image and a non-dusty reference image associated with the dusty image.

[0018] The anomaly detection method for coal mines provided in this embodiment can be applied to scenarios where cameras arranged in any area underground in a coal mine perform anomaly detection after collecting images, such as anomaly detection of belt conveyors, anomaly detection of large pieces of coal gangue, anomaly detection of discarded oil drums, anomaly detection of safety helmet wearing, etc., to eliminate the dust problem of collected images during the detection process and improve the safety of underground coal mine operations.

[0019] This embodiment takes the anomaly detection method in a coal mine in a safety helmet wearing scenario as an example to illustrate the anomaly detection method in a coal mine. The descriptions of other scenarios can refer to the same or corresponding descriptions in this embodiment, and this embodiment will not be elaborated on here.

[0020] Specifically, the image acquisition device is a camera placed in the coal mine for safety inspection. The dusty image specifically refers to the image captured by the image acquisition device at any time. It can be understood that due to the dust problem in the coal mine operation, the raised coal particles will be adsorbed on the camera, which will cause the image captured by the camera to be unclear. See Figure 2 The schematic diagrams shown in (1) and (2) are Figure 2 (1) is the real image of the corresponding camera position in the coal mine, and Figure 2 (2) is the image captured by the camera in the coal mine. The camera in the coal mine is inevitably blocked by coal dust, which will cause the captured image to be unclear. If the unclear image is used for subsequent anomaly detection, it will seriously affect the clarity. Therefore, it is necessary to remove the dust effect from the collected dusty image. It can be understood as adjusting the dusty image to an undusted state, thereby improving the detection accuracy.

[0021] Accordingly, the dust reference image specifically refers to an image with dust effects that is referenced when performing dust effect removal on a dust image, and this image is obtained from historical images captured by an image acquisition device. Accordingly, the non-dust reference image specifically refers to an image without dust effects that is referenced when performing dust effect removal on a dust image, and this image is obtained from historical images captured by an image acquisition device. In practical applications, the dust reference image and the non-dust reference image can be pre-set so that they can be directly reused when performing dust effect removal on dust images captured at any time. Alternatively, each time a dust image is captured, a selection can be made from the set corresponding to the image acquisition device according to a set of rules to further improve the accuracy of dust effect removal. This embodiment does not impose any further limitations on this.

[0022] The anomaly detection method for underground coal mines provided by the present embodiment can, in order to eliminate the dust problem existing in the images captured by the image acquisition device, first determine the dust reference image and the non-dust reference image associated with the dust image after obtaining the dust image captured by the image acquisition device arranged underground in the coal mine, so that the dust reference image and the non-dust reference image can be used to characterize the difference before and after dusting. Therefore, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined, and the dust elimination feature is calculated on this basis. The dust elimination feature can eliminate the influence of dust on the dust image, so that the dust elimination feature can be used to update the dust image, and the target image is obtained according to the update result; the target image is the image with the dust elimination effect, that is, the target image is closer to the actual situation in the real coal mine. Thereafter, the target image can be input into the anomaly detection model for processing to obtain the anomaly detection result of the image acquisition device corresponding to the underground coal mine area. After the image acquisition device captures the image but before performing anomaly detection, the collected image can be dust-removed, which can make the image after the dust effect is eliminated closer to the real situation, thereby effectively improving the accuracy of subsequent anomaly detection in coal mines, making underground coal mine operations safer and more reliable.

[0023] In one or more implementations provided in this embodiment, determining the dust reference image and the non-dust reference image associated with the dust image includes: Determine an image set associated with the image acquisition device, and determine the equipment maintenance time based on the maintenance log of the image acquisition device; select an image in the image set that has a preceding adjacent relationship with the equipment maintenance time as a dusty reference image, and select an image in the image set that has a following adjacent relationship with the equipment maintenance time as a non-dust reference image; wherein the dusty reference image and the non-dust reference image have a time-aligned relationship.

[0024] Specifically, the image collection specifically refers to a collection of historical images collected by the image acquisition device. Correspondingly, the maintenance log specifically refers to a log that records maintenance and cleaning information for the image acquisition device. The equipment maintenance time can be determined based on the log. The equipment maintenance time specifically refers to the time for regularly cleaning the image acquisition components of the image acquisition device.

[0025] Based on this, considering that the image collection associated with the image acquisition device contains a large number of historical images, in order to be able to select an image that can eliminate the dust effect of the dust image collected at the current moment, the image collection associated with the image acquisition device can be first determined, and the equipment maintenance time can be determined according to the maintenance log of the image acquisition device; the image that has a previous adjacent relationship with the equipment maintenance time is selected in the image collection as the dust reference image, and the image that has a subsequent adjacent relationship with the equipment maintenance time is selected as the non-dust reference image. It can be understood that, in the image collection, the last image collected when the image acquisition device was not cleaned before the equipment maintenance time is selected as the dust reference image, and the image that has a time-aligned relationship with the dust reference image after the image acquisition device was cleaned after the equipment maintenance time is selected as the non-dust reference image. By controlling the time alignment of the two images, it can be ensured that the two images are images before and after cleaning collected at the same time on two days, thereby eliminating the influence of other factors in the coal mine, such as avoiding the influence of the light intensity of the lighting equipment in the coal mine at different time periods, thereby making the subsequent dust effect elimination more effective.

[0026] In practical applications, considering that the image collection may contain a large number of images, even spanning multiple equipment maintenance times, in order to select a reference image that is more suitable for current use, the most recent equipment maintenance time can be selected as the time based on which the dusty reference image and the non-dusty reference image are based, thereby improving the correlation between images for subsequent use.

[0027] For example, the camera placed in area A of the coal mine collects the following Figure 2 After the dusty image shown in (2) is taken, the dusty image is covered by coal dust on the camera, resulting in an unclear image. At this time, the dusty image needs to be processed to remove the dust effect. In this process, the image set consisting of the historical images collected by the camera can be determined, and the time T when the camera was cleared can be determined. Then, the images collected before time T can be selected from the image set, such as Figure 2 As shown in (3), the dust reference image is used as the image, and the image collected after time T and in the same time interval as the dust reference image is selected, as shown in Figure 2 As shown in (4), the non-dust reference image is used as the dust-free reference image. Subsequently, the dust-free reference image and the non-dust-free reference image can be combined to complete the dust effect removal process on the dust-free image.

[0028] In summary, by selecting reference images according to the time relationship, the correlation of the images in the time dimension can be improved, thereby ensuring a stronger correlation between the reference image and the dusty image, thereby effectively improving the subsequent image processing effect.

[0029] Step S104 : determining the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image, and calculating a dust removal feature according to the dust global pixel value and the non-dust global pixel value.

[0030] Specifically, after obtaining the dust reference image and the non-dust reference image as mentioned above, in order to achieve the purpose of eliminating the dust effect, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined. The global pixel value can reflect the overall performance of the dust reference image and the non-dust reference image, and the overall performance includes the dust effect and the non-dust effect. Therefore, by combining the global pixel values corresponding to the two, the dust elimination feature can be calculated. The dust elimination feature can characterize the pixel value impact brought by the dust effect. The dust image can be updated based on this to achieve the purpose of eliminating the dust effect.

[0031] The "dust global pixel value" specifically refers to a global pixel value representing the presence of dust, calculated based on the pixel value corresponding to each pixel in the dust reference image. The "undust global pixel value" specifically refers to a global pixel value representing the absence of dust, calculated based on the pixel value corresponding to each pixel in the undust reference image. Correspondingly, the "dust removal feature" specifically refers to information about the impact of dust on pixel values in an image. By using this dust removal feature to update pixel values in an image with dust, an image free of dust can be obtained.

[0032] In one or more implementations provided in this embodiment, determining the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image, and calculating the dust removal feature based on the dust global pixel value and the non-dust global pixel value includes: The method comprises the steps of: reading a dust reference pixel value of each pixel in the dust reference image, performing a weighted average calculation on the dust reference pixel values, and obtaining a dust global pixel value of the dust reference image; reading a non-dust reference pixel value of each pixel in the non-dust reference image, performing a weighted average calculation on the non-dust reference pixel values, and obtaining a non-dust global pixel value of the non-dust reference image; performing a subtraction calculation on the dust global pixel value and the non-dust global pixel value, and determining a dust removal feature based on the calculation result.

[0033] Specifically, the weight corresponding to the dust reference pixel value can be set according to actual needs, and this embodiment does not impose any restrictions on this. Based on this, when calculating the dust removal feature, the dust reference pixel value of each pixel in the dust reference image can be first read. At this time, a weighted average calculation can be performed on the dust reference pixel value, and then the dust global pixel value of the dust reference image can be obtained based on the weighted calculation result; similarly, the non-dust reference pixel value of each pixel in the non-dust reference image can be read. At this time, a weighted average calculation can be performed on the non-dust reference pixel value, and then the non-dust global pixel value of the non-dust reference image can be obtained based on the weighted calculation result; finally, by subtracting the dust global pixel value from the non-dust global pixel value, the dust removal feature can be determined based on the calculation result for subsequent use.

[0034] In practical applications, in addition to calculating the dust removal features by subtraction as mentioned above, the dust removal features can also be determined by weighted average calculation; and, when performing weighted calculation based on the non-dusted global pixel value and the dusty global pixel value, the weight corresponding to the dusty global pixel value should be greater than the weight corresponding to the non-dusted global pixel value, so as to improve the manifestation of the dust effect, thereby making the dust removal features obtained after the weighted calculation more accurate.

[0035] In summary, by using the weighted average calculation method to construct the global pixel value corresponding to each reference image and calculating the dust removal feature based on this, the calculation result can be made more accurate, thereby improving the accuracy of subsequent dust effect removal.

[0036] In one or more implementations provided in this embodiment, on the other hand, calculating the dust removal feature according to the dust global pixel value and the non-dust global pixel value includes: An initial dust removal pixel value is calculated based on the dust global pixel value and the non-dust global pixel value; the non-dust reference image is updated using the initial dust removal pixel value to obtain a candidate dust reference image; image similarity between the candidate dust reference image and the dust reference image is calculated; when the image similarity is greater than a preset similarity threshold, the initial dust removal pixel value is used as a dust removal feature; when the image similarity is less than or equal to a preset similarity threshold, the initial dust removal pixel value is optimized using the preset similarity threshold as a constraint condition, and the optimized initial dust removal pixel value that meets the image processing condition is used as the dust removal feature.

[0037] Specifically, the initial dust removal pixel value refers to the pixel value obtained by subtracting the dust global pixel value from the non-dust global pixel value and taking the absolute value. Accordingly, the candidate dust reference image refers to the reference image obtained by updating the non-dust reference image using the initial dust removal pixel value. This can be understood as adding the initial dust removal pixel value to the non-dust reference image to achieve a dust effect, and the resulting image is the candidate dust reference image with the added dust effect. Accordingly, image similarity refers to the image similarity between the candidate dust reference image and the dust reference image; a higher similarity indicates closer images. The image processing condition specifically refers to the detection condition where the optimized initial dust removal pixel value is greater than a preset similarity threshold.

[0038] Based on this, when calculating the dust removal feature, in order to ensure more accurate calculation results, an initial dust removal pixel value can be calculated based on the dust global pixel value and the non-dust global pixel value. The calculated initial dust removal pixel value can then be tested. During the test process, the initial dust removal pixel value can be used to update the non-dust reference image, thereby adding a dust effect to the non-dust reference image to obtain a candidate dust reference image. In theory, the candidate dust reference image with the added dust effect should be the same as or similar to the dust reference image. However, since the initial dust removal pixel value is obtained by combining the above calculation method, a test is required. The image similarity between the candidate dust reference image and the dust reference image can be calculated.

[0039] When the image similarity is greater than the preset similarity threshold, it means that the candidate dust reference image is relatively similar to the dust reference image, which further indicates that the initial dust removal pixel value currently calculated is relatively accurate and can be used for subsequent dust removal effect processing of the dust image. Therefore, the initial dust removal pixel value can be used as the dust removal feature.

[0040] When the image similarity is less than or equal to the preset similarity threshold, it means that the initial dust removal pixel value calculated at this time cannot accurately represent the dust effect and needs to be optimized. At this time, the preset similarity threshold can be used as a constraint condition to optimize the initial dust removal pixel value. During the optimization process, the optimized initial dust removal pixel value can be fused with the non-dusted reference image at each optimization stage, and the similarity between the fused image and the dust reference image can be calculated, and then compared with the similarity threshold. This is iterated until the initial dust removal pixel value that meets the image processing conditions is determined, which can be used as the dust removal feature for subsequent use.

[0041] In addition, after obtaining the initial dust removal pixel value, the initial dust removal pixel value can also be used to update the dust reference image to obtain a candidate non-dust reference image with the dust effect removed. Thereafter, the image similarity between the candidate non-dust reference image and the non-dust reference image can be calculated; when the image similarity is greater than a preset similarity threshold, the initial dust removal pixel value is used as a dust removal feature; when the image similarity is less than or equal to a preset similarity threshold, the preset similarity threshold is used as a constraint condition to optimize the initial dust removal pixel value, and the optimized initial dust removal pixel value that meets the image processing conditions is used as the dust removal feature.

[0042] During specific implementation, the method for calculating the dust removal feature can be selected according to actual needs, and this embodiment does not impose any limitation thereto.

[0043] Using the above example, we can get Figure 2 After obtaining the dust reference image and the non-dust reference image shown in (3) and (4), the global pixel value Q1 corresponding to the dust reference image and the global pixel value Q2 corresponding to the non-dust reference image can be determined by weighted average calculation. The global pixel value Q1 contains information about the impact of the dust effect on the pixel value. Therefore, by calculating |global pixel value Q1-global pixel value Q2|, the initial dust removal pixel value q can be obtained.

[0044] Furthermore, to verify the usability of the initial dust-removed pixel value q, the non-dust reference image can be updated using the initial dust-removed pixel value q. Based on the updated result, an image incorporating the dust effect corresponding to the initial dust-removed pixel value q can be obtained. The image similarity between this image and the dust-removed reference image can then be calculated. If the image similarity is greater than a preset similarity threshold M, it indicates that the initial dust-removed pixel value q can fully represent the dust effect and can therefore be used as the dust-removed pixel value Q for subsequent use. If the image similarity is less than or equal to the preset similarity threshold M, it indicates that the initial dust-removed pixel value q cannot fully represent the dust effect. Therefore, the preset similarity threshold M can be used as a constraint to optimize the initial dust-removed pixel value q until the dust-removed pixel value Q is obtained, and then the dust image can be processed.

[0045] In summary, by verifying the calculated dust removal pixel value, it can be ensured that the pixel value can carry the pixel value influence information corresponding to the dust effect, which can be used for subsequent image updates. It can ensure that the updated image is closer to the real image, thereby improving the accuracy of anomaly detection.

[0046] Step S106 : updating the dust image using the dust removal feature, and obtaining a target image according to the update result.

[0047] Specifically, after obtaining the dust removal feature as described above, considering that the dust removal feature corresponds to the pixel value influence of the dust effect, the dust removal feature is used to update the dust image, so that the dust effect can be eliminated in the dust image, thereby obtaining the target image according to the update result, so as to complete the anomaly detection in the underground area of the coal mine based on the target image in the subsequent process.

[0048] The update operation can be understood as uniformly subtracting or adding the dust removal feature from the pixel values of the dust image to obtain the target image, achieving the purpose of removing the dust effect. It should be noted that pixel values typically range from 0 to 255. If the subtraction or addition of the dust removal feature causes the pixel value to fall outside this range, 0 or 255 can be directly selected as the corresponding pixel value to avoid image display errors. The target image is the image obtained after removing the dust effect.

[0049] In one or more implementations provided in this embodiment, updating the dust image using the dust removal feature and obtaining the target image according to the update result includes: A grayscale curve corresponding to the dust image is constructed, and trend detection is performed on the grayscale curve; when it is determined based on the trend detection result that a local fluctuation curve exists in the grayscale curve, a non-dust area is determined in the dust image based on the local passive curve; an image area in the dust image other than the non-dust area is used as a dust area, and the dust area is updated using the dust removal feature, and a target image is obtained based on the update result.

[0050] Specifically, a grayscale curve refers to a curve plotted based on the grayscale value corresponding to each pixel in a dusty image. This curve can reflect the grayscale value trend for each pixel in the dusty image. Given the impact of coal dust on image acquisition equipment, a grayscale curve should typically be smooth and continuous. However, it is possible that the image acquisition equipment may be affected by external factors, with some areas being significantly affected by dust and others less so. This can cause significant variations in grayscale values across different regions of the dusty image, resulting in peaks and troughs in the grayscale curve. For example, if an operator accidentally touches the camera, the touch may cause a portion of the camera to become less dusty. In this case, dust removal can be performed to ignore this region, thus avoiding distortion after the overall image is dust-removed. Accordingly, the local fluctuation curve corresponds to regions with less dust. Non-dusty areas are those that do not require dust removal, while dusty areas are those that do.

[0051] Based on this, when performing dust removal, a grayscale curve corresponding to the dusty image can be constructed and trend detection can be performed on this grayscale curve. This trend detection can determine the uniformity of dust accumulation on the image acquisition device. If the trend detection results indicate that there are local fluctuations in the grayscale curve, this indicates that the image acquisition device is unevenly dusted. This indicates that dust removal processing does not need to be performed on the entire image, but only on certain areas. Therefore, based on this local passive curve, non-dust areas can be identified in the dusty image. The image areas outside of these non-dust areas can then be identified as dusty areas, and the dust removal features can be used to update these areas. The target image can then be obtained based on the updated results.

[0052] In addition, when it is determined based on the trend detection results that there is no local fluctuation curve in the grayscale curve, the dusty image as a whole can be directly processed to remove the dust effect, that is, the dusty image can be directly updated using the dust removal feature, so as to obtain the target image based on the update result for subsequent use.

[0053] In specific implementation, how to determine whether there is a local fluctuation curve in the grayscale curve can be determined by calculating the distance between the peak and the trough. When the distance between the peak and the trough is greater than the preset distance threshold, it can be determined that there is a local fluctuation curve in the grayscale curve, otherwise it is not, thereby avoiding triggering the above calculation logic when the fluctuation is small, which leads to waste of computing resources.

[0054] In summary, by using the grayscale curve trend detection method to determine whether the dusty image needs to be dust-removed as a whole, the update operation can be completed dynamically, thereby effectively improving the utilization of computing resources.

[0055] In one or more implementations provided in this embodiment, after the step of updating the dust image using the dust removal feature and obtaining the target image according to the update result is performed, the method further includes: The steps of calculating the structural similarity between the target image and the dust image, and extracting dust feature points and target feature points from the dust image and the target image using a feature extraction algorithm; calculating the feature point matching between the dust feature points and the target feature points; and inputting the target image into an anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine when determining that the target image and the dust image meet the detection conditions based on the structural similarity and the feature point matching.

[0056] Specifically, structural similarity refers to the similarity calculated using SSIM (Structural Similarity Index), which can reflect the similarity between images in terms of brightness, contrast, and structure. Dust feature points and target feature points specifically refer to pixel points with a symmetrical relationship on the dust image and the target image. The feature point matching degree is the value that reflects the degree of matching between the two images.

[0057] Based on this, after the dust image is updated using the dust removal feature, the target image can be obtained. In order to verify the effect of dust removal, the structural similarity between the target image and the dust image can be calculated, and the feature extraction algorithm can be used to extract dust feature points and target feature points in the dust image and the target image, and the feature point matching between the dust feature points and the target feature points can be calculated; thereafter, the target image can be verified based on the structural similarity and feature point matching. When it is determined that the target image and the dust image meet the detection conditions based on the structural similarity and feature point matching, it can be said that the dust removal effect of the target image is relatively good, so the step of inputting the target image into the anomaly detection model for processing can be executed to obtain the anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine.

[0058] Using the above example, after obtaining the dust removal pixel value Q, we can Figure 2 On the dust image shown in (2), the dust elimination pixel value Q is subtracted from the pixel value corresponding to each pixel. The calculation results show that Figure 2 The target image shown in (5) achieves the purpose of eliminating the dust effect. Furthermore, in order to verify the elimination result of the dust effect, the structural similarity between the dust image and the target image can be calculated, and the feature point matching degree can be calculated by extracting feature points. When the structural similarity is greater than the preset structural similarity threshold and the feature point matching degree is greater than the preset feature point matching degree threshold, it can be determined that the target image is a high-quality image with the dust effect eliminated. Subsequently, anomaly detection in the coal mine can be performed based on the target image.

[0059] In summary, by verifying the images with the dust effect eliminated, it is possible to ensure that high-quality images are subsequently used for anomaly detection, thereby effectively improving the accuracy of anomaly detection.

[0060] Step S108: input the target image into an anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground coal mine area.

[0061] Specifically, after obtaining a high-quality target image as described above, the target image can be input into an anomaly detection model for processing. The anomaly detection model can predict anomaly detection results for the underground coal mine area corresponding to the image acquisition device, allowing downstream businesses to use these anomaly detection results to make alert decisions. Specifically, the anomaly detection model refers to a large language model that inputs a target image and outputs anomaly detection results. Based on pre-programmed prompts within the model, it can detect anomalies in the area corresponding to the target image. For example, it can detect whether workers are wearing helmets, whether there are large pieces of coal gangue on the conveyor belt, whether there are oil leaks in oil drums, or whether the conveyor belt is broken. Accordingly, the underground coal mine area corresponds to the area corresponding to the image acquisition device. In specific implementations, the anomaly detection model can be implemented using a large language model, which needs to be fine-tuned based on the business scenario to ensure that the model has the ability to detect anomalies in underground coal mines.

[0062] In one or more implementations provided in this embodiment, after the step of inputting the target image into the anomaly detection model for processing to obtain an anomaly detection result corresponding to the underground area of the coal mine by the image acquisition device is performed, the following steps are further included: In the case that the abnormality detection result fails, an alarm message is constructed for the underground area of the coal mine according to the abnormality detection result, and the alarm message is sent to the operation and maintenance terminal; in the case that the abnormality detection result passes, a model optimization sample pair is constructed based on the abnormality detection result, the dusty image and the target image, and the model optimization sample pair is stored in a sample set, wherein the model optimization sample pair contained in the sample set is used to optimize the abnormality detection model.

[0063] Specifically, alert information refers to the alert information sent to the operation and maintenance terminal based on the anomaly detection results. It is used to inform the operation and maintenance personnel of the current problem, so that the anomaly problem can be quickly resolved to avoid excessive losses. Correspondingly, model optimization sample pairs refer to the sample pairs used to optimize the anomaly detection model. They are used to achieve the goal of optimizing the model during use, ensuring that the model always has strong predictive capabilities.

[0064] Based on this, after using the anomaly detection model to determine the anomaly detection results corresponding to the underground coal mine area, if the anomaly detection result fails, it indicates that there is a problem in the current area. Therefore, an alarm message can be generated for the underground coal mine area based on the anomaly detection result and sent to the operation and maintenance terminal. If the anomaly detection result passes, a model optimization sample pair can be constructed based on the anomaly detection result, the dust image, and the target image. The model optimization sample pair is stored in the sample collection, so that the model optimization sample pair contained in the sample collection can be used to optimize the anomaly detection model later.

[0065] Using the above example, we can get Figure 2 After obtaining the target image shown in (5), the target image can be input into the anomaly detection model for processing. According to the processing result, it is determined that {there are workers who are not wearing helmets in area A}. At this time, an alarm message can be sent to the operation and maintenance terminal so that the operation and maintenance personnel at the operation and maintenance terminal can promptly notify the workers to wear helmets, thereby improving the safety of the operation.

[0066] The anomaly detection method for underground coal mines provided by the present embodiment can, in order to eliminate the dust problem existing in the images captured by the image acquisition device, first determine the dust reference image and the non-dust reference image associated with the dust image after obtaining the dust image captured by the image acquisition device arranged underground in the coal mine, so that the dust reference image and the non-dust reference image can be used to characterize the difference before and after dusting. Therefore, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined, and the dust elimination feature is calculated on this basis. The dust elimination feature can eliminate the influence of dust on the dust image, so that the dust elimination feature can be used to update the dust image, and the target image is obtained according to the update result; the target image is the image with the dust elimination effect, that is, the target image is closer to the actual situation in the real coal mine. Thereafter, the target image can be input into the anomaly detection model for processing to obtain the anomaly detection result of the image acquisition device corresponding to the underground coal mine area. After the image acquisition device captures the image but before performing anomaly detection, the collected image can be dust-removed, which can make the image after the dust effect is eliminated closer to the real situation, thereby effectively improving the accuracy of subsequent anomaly detection in coal mines, making underground coal mine operations safer and more reliable.

[0067] The following combined Figure 3 , taking the application of the anomaly detection method in a dangerous warning scenario provided in this specification as an example, the anomaly detection method in a coal mine is further described. Figure 3 A flowchart of a processing process of an anomaly detection method in a coal mine provided by an embodiment of this specification is shown, which specifically includes the following steps.

[0068] Step S302: obtaining a dust image captured by an image capture device arranged underground in the coal mine.

[0069] Step S304: determining an image set associated with the image acquisition device, and determining a device maintenance time according to a maintenance log of the image acquisition device.

[0070] Step S306 : Select an image in the image set that is adjacent to the equipment maintenance time as a dusty reference image, and select an image that is adjacent to the equipment maintenance time as a non-dusty reference image.

[0071] Step S308 : reading the dust reference pixel value of each pixel point in the dust reference image, performing weighted average calculation on the dust reference pixel values, and obtaining a global dust pixel value of the dust reference image.

[0072] Step S310 , reading the non-dust reference pixel value of each pixel in the non-dust reference image, performing weighted average calculation on the non-dust reference pixel values, and obtaining a non-dust global pixel value of the non-dust reference image.

[0073] Step S312: Calculate an initial dust-removed pixel value based on the dust global pixel value and the non-dust global pixel value.

[0074] Step S314 : updating the non-dust reference image using the initial dust-removed pixel values to obtain a candidate dust reference image.

[0075] Step S316 : Calculate the image similarity between the candidate dust reference image and the dust reference image.

[0076] Step S318 : When the image similarity is greater than a preset similarity threshold, the initial dust removal pixel value is used as a dust removal feature.

[0077] Step S320 , when the image similarity is less than or equal to a preset similarity threshold, the preset similarity threshold is used as a constraint condition to optimize the initial dust removal pixel value, and the optimized initial dust removal pixel value that meets the image processing condition is used as a dust removal feature.

[0078] Step S322: construct a grayscale curve corresponding to the dusty image, and perform trend detection on the grayscale curve.

[0079] Step S324 : When it is determined according to the trend detection result that there is a local fluctuation curve in the grayscale curve, a non-dust area is determined in the dusty image according to the local passive curve.

[0080] Step S326 : The image area except the dust area in the dust image is used as the dust area, and the dust area is updated using the dust removal feature, and the target image is obtained according to the update result.

[0081] Step S328: Input the target image into the anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine.

[0082] Step S330: If the abnormality detection result fails, alarm information is constructed for the underground area of the coal mine according to the abnormality detection result, and the alarm information is sent to the operation and maintenance terminal.

[0083] To sum up, in order to eliminate the dust problem in the images captured by the image acquisition device, after obtaining the dust image captured by the image acquisition device arranged underground in the coal mine, the dust reference image and the non-dust reference image associated with the dust image can be determined first, so that the difference before and after the dust can be represented by the dust reference image and the non-dust reference image. Therefore, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined, and the dust elimination feature is calculated on this basis. The dust elimination feature can eliminate the influence of dust on the dust image, so as to realize the update of the dust image by using the dust elimination feature, and obtain the target image according to the update result; the target image is the image with the dust elimination effect, that is, the target image is closer to the actual situation underground in the real coal mine, and then the target image can be input into the anomaly detection model for processing to obtain the anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine. After the image acquisition device captures the image but before performing anomaly detection, the collected image can be dust-removed, which can make the image after the dust effect is eliminated closer to the real situation, thereby effectively improving the accuracy of subsequent anomaly detection in coal mines, making underground coal mine operations safer and more reliable.

[0084] Corresponding to the above method embodiment, this specification also provides an embodiment of an abnormality detection device in a coal mine. Figure 4 FIG1 shows a schematic diagram of the structure of an abnormality detection device for a coal mine provided by an embodiment of this specification. Figure 4 As shown, the device includes: An acquisition module 402 is configured to acquire a dusty image acquired by an image acquisition device arranged underground in a coal mine, and determine a dusty reference image and a non-dusty reference image associated with the dusty image; a determination module 404 configured to determine a dust global pixel value of the dust reference image and a non-dust global pixel value of the non-dust reference image, and calculate a dust removal feature based on the dust global pixel value and the non-dust global pixel value; An updating module 406 is configured to update the dust image using the dust removal feature and obtain a target image according to the updating result; The processing module 408 is configured to input the target image into the anomaly detection model for processing, and obtain an anomaly detection result of the image acquisition device corresponding to the underground coal mine area.

[0085] In an optional embodiment, determining the dust reference image and the non-dust reference image associated with the dust image includes: Determine an image set associated with the image acquisition device, and determine the equipment maintenance time based on the maintenance log of the image acquisition device; select an image in the image set that has a preceding adjacent relationship with the equipment maintenance time as a dusty reference image, and select an image in the image set that has a following adjacent relationship with the equipment maintenance time as a non-dust reference image; wherein the dusty reference image and the non-dust reference image have a time-aligned relationship.

[0086] In an optional embodiment, determining the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image, and calculating the dust removal feature based on the dust global pixel value and the non-dust global pixel value includes: The method comprises the steps of: reading a dust reference pixel value of each pixel in the dust reference image, performing a weighted average calculation on the dust reference pixel values, and obtaining a dust global pixel value of the dust reference image; reading a non-dust reference pixel value of each pixel in the non-dust reference image, performing a weighted average calculation on the non-dust reference pixel values, and obtaining a non-dust global pixel value of the non-dust reference image; performing a subtraction calculation on the dust global pixel value and the non-dust global pixel value, and determining a dust removal feature based on the calculation result.

[0087] In an optional embodiment, the calculating of the dust removal feature according to the dust global pixel value and the non-dust global pixel value includes: An initial dust removal pixel value is calculated based on the dust global pixel value and the non-dust global pixel value; the non-dust reference image is updated using the initial dust removal pixel value to obtain a candidate dust reference image; image similarity between the candidate dust reference image and the dust reference image is calculated; when the image similarity is greater than a preset similarity threshold, the initial dust removal pixel value is used as a dust removal feature; when the image similarity is less than or equal to a preset similarity threshold, the initial dust removal pixel value is optimized using the preset similarity threshold as a constraint condition, and the optimized initial dust removal pixel value that meets the image processing condition is used as the dust removal feature.

[0088] In an optional embodiment, updating the dust image using the dust removal feature and obtaining the target image according to the update result includes: A grayscale curve corresponding to the dust image is constructed, and trend detection is performed on the grayscale curve; when it is determined based on the trend detection result that a local fluctuation curve exists in the grayscale curve, a non-dust area is determined in the dust image based on the local passive curve; an image area in the dust image other than the non-dust area is used as a dust area, and the dust area is updated using the dust removal feature, and a target image is obtained based on the update result.

[0089] In an optional embodiment, after the step of inputting the target image into an anomaly detection model for processing and obtaining an anomaly detection result of the image acquisition device corresponding to the underground coal mine area is performed, the method further includes: In the case that the abnormality detection result fails, an alarm message is constructed for the underground area of the coal mine according to the abnormality detection result, and the alarm message is sent to the operation and maintenance terminal; in the case that the abnormality detection result passes, a model optimization sample pair is constructed based on the abnormality detection result, the dusty image and the target image, and the model optimization sample pair is stored in a sample set, wherein the model optimization sample pair contained in the sample set is used to optimize the abnormality detection model.

[0090] In an optional embodiment, after the step of updating the dust image using the dust removal feature and obtaining the target image according to the update result is performed, the method further includes: The steps of calculating the structural similarity between the target image and the dust image, and extracting dust feature points and target feature points from the dust image and the target image using a feature extraction algorithm; calculating the feature point matching between the dust feature points and the target feature points; and inputting the target image into an anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine when determining that the target image and the dust image meet the detection conditions based on the structural similarity and the feature point matching.

[0091] The anomaly detection device for underground coal mines provided by the present embodiment can, in order to eliminate the dust problem existing in images captured by the image acquisition device, first determine the dust reference image and the non-dust reference image associated with the dust image after obtaining the dust image captured by the image acquisition device arranged underground in the coal mine, so that the dust reference image and the non-dust reference image can represent the difference before and after dusting. Therefore, the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image can be determined, and the dust elimination feature is calculated based on this. The dust elimination feature can eliminate the influence of dust on the dust image, so that the dust elimination feature can be used to update the dust image, and the target image is obtained according to the update result; the target image is the image with the dust elimination effect, that is, the target image is closer to the actual situation in the real coal mine. Thereafter, the target image can be input into the anomaly detection model for processing to obtain the anomaly detection result of the image acquisition device corresponding to the underground coal mine area. After the image acquisition device captures the image but before performing anomaly detection, the collected image can be dust-removed, which can make the image after the dust effect is eliminated closer to the real situation, thereby effectively improving the accuracy of subsequent anomaly detection in coal mines, making underground coal mine operations safer and more reliable.

[0092] The above is a schematic diagram of an anomaly detection device for an underground coal mine according to this embodiment. It should be noted that the technical solution of this underground coal mine anomaly detection device and the technical solution of the underground coal mine anomaly detection method described above are based on the same concept. For details not described in detail in the technical solution of the underground coal mine anomaly detection device, please refer to the description of the technical solution of the underground coal mine anomaly detection method described above.

[0093] Figure 5 The block diagram of a computing device 500 according to one embodiment of the present disclosure is shown. Components of the computing device 500 include, but are not limited to, a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and a database 550 is used to store data.

[0094] The computing device 500 also includes an access device 540 that enables the computing device 500 to communicate via one or more networks 560. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 540 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, or a near field communication (NFC) interface.

[0095] In one embodiment of the present specification, the above components of the computing device 500 and Figure 5 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 5 The computing device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art may add or replace other components as needed.

[0096] Computing device 500 can be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or personal computer (PC). Computing device 500 can also be a mobile or stationary server.

[0097] The processor 520 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-mentioned anomaly detection method in underground coal mines.

[0098] The above is a schematic diagram of a computing device according to this embodiment. It should be noted that the technical solution of this computing device is based on the same concept as the technical solution of the above-mentioned coal mine anomaly detection method. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the above-mentioned coal mine anomaly detection method.

[0099] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the above-mentioned anomaly detection method in an underground coal mine.

[0100] The above is a schematic diagram of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium is based on the same concept as the technical solution of the above-mentioned coal mine anomaly detection method. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the above-mentioned coal mine anomaly detection method.

[0101] An embodiment of the present specification further provides a computer program product, including a computer program or instructions, which, when executed by a processor, implements the steps of the above-mentioned anomaly detection method in a coal mine.

[0102] The above is a schematic diagram of a computer program product according to this embodiment. It should be noted that the technical solution of this computer program product is based on the same concept as the technical solution of the above-mentioned coal mine anomaly detection method. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the above-mentioned coal mine anomaly detection method.

[0103] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0104] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium may be appropriately increased or decreased based on the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0105] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0106] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0107] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of the embodiments described herein. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification.

Claims

1. A method for detecting anomalies in a coal mine, characterized in that: include: Acquire a dusty image captured by an image capture device arranged underground in a coal mine, and determine a dusty reference image and a non-dusty reference image associated with the dusty image; determining a dust global pixel value of the dust reference image and a non-dust global pixel value of the non-dust reference image, and calculating a dust removal feature based on the dust global pixel value and the non-dust global pixel value; updating the dust image using the dust removal feature, and obtaining a target image according to the update result; The target image is input into an anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground coal mine area.

2. The anomaly detection method in a coal mine according to claim 1, characterized in that: The determining of the dust reference image and the non-dust reference image associated with the dust image includes: Determining an image set associated with the image acquisition device, and determining a device maintenance time based on a maintenance log of the image acquisition device; Selecting, from the image set, an image that is adjacent to the device maintenance time before the image is used as a dusty reference image, and selecting an image that is adjacent to the device maintenance time after the image is used as a non-dusty reference image; The dusty reference image and the non-dusty reference image have a time alignment relationship.

3. The anomaly detection method in a coal mine according to claim 1, characterized in that: The determining of the dust global pixel value of the dust reference image and the non-dust global pixel value of the non-dust reference image, and calculating the dust removal feature according to the dust global pixel value and the non-dust global pixel value, includes: Reading a dust reference pixel value of each pixel point in the dust reference image, performing a weighted average calculation on the dust reference pixel value, and obtaining a dust global pixel value of the dust reference image; Reading a non-dust reference pixel value of each pixel point in the non-dust reference image, performing weighted average calculation on the non-dust reference pixel values, and obtaining a non-dust global pixel value of the non-dust reference image; A subtraction calculation is performed on the dusty global pixel value and the non-dusted global pixel value, and a dust removal feature is determined according to the calculation result.

4. The anomaly detection method in a coal mine according to claim 1, characterized in that: The calculating of the dust removal feature according to the dust global pixel value and the non-dust global pixel value includes: Calculating an initial dust-removed pixel value according to the dust-removed global pixel value and the non-dust-removed global pixel value; updating the non-dust reference image using the initial dust-removed pixel values to obtain a candidate dust reference image; Calculating image similarity between the candidate dust reference image and the dust reference image; When the image similarity is greater than a preset similarity threshold, using the initial dust removal pixel value as a dust removal feature; When the image similarity is less than or equal to a preset similarity threshold, the initial dust removal pixel value is optimized with the preset similarity threshold as a constraint condition, and the optimized initial dust removal pixel value that meets the image processing condition is used as a dust removal feature.

5. The anomaly detection method in a coal mine according to claim 1, characterized in that: The updating of the dust image by using the dust removal feature and obtaining a target image according to the updating result includes: constructing a grayscale curve corresponding to the dusty image, and performing trend detection on the grayscale curve; When it is determined according to the trend detection result that there is a local fluctuation curve in the grayscale curve, determining a non-dust area in the dust image according to the local passive curve; An image area in the dust image excluding the non-dust area is used as a dust area, and the dust area is updated using the dust removal feature, and a target image is obtained according to the update result.

6. The method for detecting anomalies in a coal mine according to any one of claims 1 to 5, characterized in that: After the step of inputting the target image into the anomaly detection model for processing to obtain an anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine is performed, the method further includes: If the abnormality detection result fails, constructing alarm information for the underground area of the coal mine according to the abnormality detection result, and sending the alarm information to the operation and maintenance terminal; When the anomaly detection result passes, a model optimization sample pair is constructed based on the anomaly detection result, the dusty image and the target image, and the model optimization sample pair is stored in a sample set, wherein the model optimization sample pair contained in the sample set is used to optimize the anomaly detection model.

7. The method for detecting anomalies in an underground coal mine according to any one of claims 1 to 5, characterized in that: After the step of updating the dust image by using the dust removal feature and obtaining the target image according to the update result is performed, the method further includes: Calculating the structural similarity between the target image and the dust image, and extracting dust feature points and target feature points from the dust image and the target image using a feature extraction algorithm; Calculating a feature point matching degree between the dust feature point and the target feature point; When it is determined that the target image and the dusty image meet the detection conditions based on the structural similarity and the feature point matching degree, the step of inputting the target image into the anomaly detection model for processing is executed to obtain the anomaly detection result of the image acquisition device corresponding to the underground area of the coal mine.

8. A computing device, characterized in that include: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium, characterized in that It stores computer-executable instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that The method comprises a computer program or instructions, which implements the steps of the method according to any one of claims 1 to 7 when executed by a processor.

Images