A method, device, equipment and storage medium for detecting gas leakage in mines based on Fourier transform

Abstract

The present invention discloses a method for detecting gas leakage in mines based on Fourier transform, which enhances the detection capability of gas leakage areas in mines by combining the semantic information of visible light images and the temperature information of infrared images. The methods such as multimodal image matching using LoFTR, deep features extracted by CNN neural network, FFT conversion to frequency domain features, FMLP neural network weighted processing and inverse Fourier transform are used to provide rich data support for image reconstruction and fusion. On this basis, an intelligent model that can accurately identify and analyze safety hazards in mining areas is constructed through model training, verification and optimization. Finally, the trained model is deployed in practical applications, realizing real-time monitoring and analysis of mining areas, effectively preventing and reducing the occurrence of mine safety accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of mine gas leakage detection, and in particular to a mine gas leakage detection method, device, equipment and storage medium based on Fourier transform. Background Art

[0002] In mining operations, the necessity of gas leak detection not only affects the safety of miners but is also crucial for ensuring stable mine operations and environmental protection. The composition of gases in underground mines is complex. With the acceleration of industrialization and the increasing frequency of mining activities, the risk of gas leaks is also increasing. Leaks of harmful gases such as methane and carbon monoxide can not only cause explosions and poisoning accidents, but also cause serious environmental pollution. Therefore, to improve mine safety management, reduce accidents, protect miners' lives, and maintain sustainable environmental development, it is necessary to detect gas leaks in mines.

[0003] However, existing methods for detecting gas leaks in mines have numerous limitations. Existing technologies, such as catalytic combustion, thermal conductivity, and optical interferometry, are often affected by environmental factors such as coal dust, temperature, humidity, and electromagnetic interference, resulting in inaccurate measurements. Furthermore, these methods are limited in their ability to provide real-time monitoring and early warning, failing to meet the rapid response and high sensitivity requirements of modern mines. Therefore, the development of a more efficient and accurate method for detecting gas leaks in mines is urgent. Summary of the Invention

[0004] The purpose of the present invention is to solve the above technical problems and provide a mine gas leakage detection method based on Fourier transform, which can improve the accuracy and real-time performance of mine gas detection.

[0005] In order to solve the above problems, the present invention is implemented according to the following technical solutions:

[0006] In a first aspect, the present invention provides a method for detecting gas leakage in a mine based on Fourier transform, the method comprising:

[0007] S100, obtaining a visible light image and an infrared image of the mine to be inspected;

[0008] S200, extracting local feature data from the visible light image and the infrared image, respectively, wherein the local feature data includes coarse-grained features and fine-grained features;

[0009] S300, extracting image feature points from the local feature data and performing multimodal image matching to obtain matching spatial position information data between the visible light image and the infrared image;

[0010] S400, extracting respective deep feature data from the visible light image and the infrared image based on the spatial position information data, wherein the deep feature data includes deep semantic information of the image;

[0011] S500, converting the deep feature data from the spatial domain to the frequency domain using a fast Fourier transform algorithm to obtain complex feature data;

[0012] S600, weighting the complex feature data through a Fourier multilayer perceptron;

[0013] S700, converting the weighted complex feature data from the frequency domain back to the spatial domain through an inverse fast Fourier transform algorithm;

[0014] S800: Constructing a mine gas leakage detection model, and training the mine gas leakage detection model using the data obtained in step S700;

[0015] S900 , performing model verification on the trained mine gas leakage detection model, and adjusting the trained mine gas leakage detection model according to the verification result to obtain a detection model that can be used to detect mine gas leakage.

[0016] Preferably, in step S200, local feature data of each is extracted from the visible light image and the infrared image respectively through a convolutional neural network.

[0017] Preferably, in step S300, image feature points are extracted from the local features using a LoFTR local feature matching algorithm.

[0018] Preferably, in step S400, based on the spatial position information data, respective deep feature data are extracted from the visible light image and the infrared image through a convolutional neural network.

[0019] Preferably, step S700 includes the following specific steps: S710, converting the weighted complex feature data from the frequency domain back to the spatial domain through an inverse fast Fourier transform algorithm to obtain fused time domain feature data; S720, performing image reconstruction through the fused time domain feature data to obtain a fused image, wherein the fused image includes the semantic information data of the visible light image and the temperature information data of the infrared image.

[0020] Preferably, step S700 further includes: S730, post-processing the fused image to improve the quality of the fused image.

[0021] Preferably, in step S900, the trained mine gas leakage detection model is validated by a cross-validation method.

[0022] In a second aspect, the present invention further provides a mine gas leakage detection device based on Fourier transform, wherein the mine gas leakage detection device based on Fourier transform is configured to execute the mine gas leakage detection method based on Fourier transform, and the mine gas leakage detection device based on Fourier transform comprises:

[0023] An image acquisition module, which is used to acquire visible light images and infrared images of the mine to be inspected;

[0024] A local feature extraction module, configured to extract local feature data from the visible light image and the infrared image, respectively, wherein the local feature data includes coarse-grained features and fine-grained features;

[0025] A multimodal image matching module, configured to extract image feature points from the local feature data and perform multimodal image matching to obtain spatial position information data of the matching between the visible light image and the infrared image;

[0026] A deep feature extraction module, configured to extract respective deep feature data from the visible light image and the infrared image based on the spatial position information data, wherein the deep feature data includes deep semantic information of the image;

[0027] A feature fusion module is configured to convert the deep feature data from the spatial domain to the frequency domain using a fast Fourier transform algorithm to obtain complex feature data; weight the complex feature data using a Fourier multilayer perceptron; and convert the weighted complex feature data from the frequency domain back to the spatial domain using an inverse fast Fourier transform algorithm;

[0028] A model building and training module, which is used to build a mine gas leakage detection model and train the mine gas leakage detection model using the data obtained in step S700;

[0029] The model verification module is used to perform model verification on the trained mine gas leakage detection model, and adjust the trained mine gas leakage detection model according to the verification result to obtain a detection model that can be used to detect mine gas leakage.

[0030] In a third aspect, the present invention provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute a mine leakage gas detection method based on Fourier transform as described in any one of claims 1 to 7.

[0031] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program is used to enable a processor to implement a mine gas leakage detection method based on Fourier transform as described in any one of claims 1 to 7 when executed.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] The present invention provides a method for detecting gas leaks in mines based on Fourier transforms. First, through multimodal image acquisition, combined with visible light and infrared imaging technologies, the semantic information of visible light images and the temperature information of infrared images are fully utilized to enhance the detection capability of leakage areas. Second, multimodal image matching is performed using LoFTR to accurately find the correspondence between the two images, laying a solid foundation for subsequent feature extraction and fusion. Next, multidimensional analysis of image features is achieved through deep features extracted by the CNN neural network and frequency domain features converted by FFT, while weighted processing by the FMLP neural network further enhances the cross-domain feature fusion effect. The inverse Fourier transform converts the fused frequency domain information back to the time domain, providing rich data support for image reconstruction and fusion. On this basis, through model training, verification, and optimization, an intelligent model capable of accurately identifying and analyzing safety hazards in mining areas is constructed. Finally, the trained model is deployed in practical applications, achieving real-time monitoring and analysis of mining areas, effectively preventing and reducing the occurrence of mine safety accidents.

[0034] Overall, the Fourier transform-based mine gas leakage detection method described in the present invention not only improves the accuracy and real-time performance of mine gas detection, but also enhances the generalization ability and adaptability of the model through the application of deep learning technology, providing strong technical support for mine safety management, and has important practical application value and broad market prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0036] Figure 1 This is a principle structure diagram of a mine gas leakage detection method based on Fourier transform of the present invention;

[0037] Figure 2 This is a technical flow chart of a preferred embodiment of a mine gas leakage detection method based on Fourier transform of the present invention;

[0038] Figure 3 This is a module diagram of a mine gas leakage detection device based on Fourier transform according to the present invention;

[0039] Figure 4 It is a structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0041] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. Unless otherwise defined, the technical or scientific terms used in this specification should have the common meaning understood by people with ordinary skills in the field to which this application belongs. The terms "first", "second" and similar terms used in this specification and claims do not indicate any order, quantity or importance, but are only used to distinguish different technical features.

[0042] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0043] like Figure 2 As shown, the method for detecting gas leakage in a mine based on Fourier transform according to the present invention comprises the following steps:

[0044] S100: Obtain visible light images and infrared images of a mine to be inspected.

[0045] Preferably, visible light images of the mine to be inspected can be obtained by installing a camera on a mining intelligent vehicle or using a drone equipped with a camera, and single-channel infrared images can be collected by an infrared night vision camera on a mining intelligent vehicle.

[0046] S200 , extracting local feature data from the visible light image and the infrared image respectively, where the local feature data includes coarse-grained features and fine-grained features.

[0047] It should be noted that local features refer to features of specific areas or points of interest (such as corners, edges, textures, etc.) in an image. These features are usually extracted by detecting specific patterns in the image, such as algorithms such as SIFT (Scale Invariant Feature Transform), SURF (Speeded Robust Features), or ORB (Oriented FAST and Rotated BRIEF). Preferably, local feature data is extracted from visible light images and infrared images respectively using convolutional neural networks. Coarse-grained features generally refer to more abstract and global information extracted from an image. They capture the overall structure and layout of the image at a larger scale. These features include, but are not limited to, the global color distribution, texture, shape, etc. of the image. They can provide an overall description of the image, but may not be rich in detailed information. Coarse-grained features are very useful for processing tasks that require global understanding, such as scene classification and object recognition. Fine-grained features refer to local features with smaller scales in the image. They usually include details and texture information in the image, such as edges, corners, local textures, etc. These features can capture more detailed aspects of an image and are particularly important for tasks requiring precise classification or recognition, such as object segmentation in an image and facial keypoint detection in face recognition. In this paper, coarse-grained features provide contextual information, while fine-grained features provide critical local and detailed information. The combination of the two forms complementary coarse-grained and fine-grained representations, enhancing the model's recognition capabilities.

[0048] Preferably, local feature data of each is extracted from the visible light image and the infrared image respectively through a convolutional neural network.

[0049] It should be noted that convolutional neural network (CNN) is a deep learning model that performs well in signal data processing in fields such as images and videos, especially in classification and recognition tasks. In the present invention, local feature data of each is extracted from visible light images and infrared images respectively by convolutional neural network, which has the following advantages: first, CNN can automatically learn and extract features from image data without manually designing or selecting feature extractors, which greatly reduces the workload of feature engineering and improves the efficiency and accuracy of feature extraction. Secondly, CNN reduces the number of network parameters and computational complexity through parameter sharing and sparse connections, making the network more efficient while reducing the risk of overfitting. Furthermore, CNN has the characteristic of translation invariance to input data, that is, the model can maintain a stable output for small changes in input data, which has good robustness for small movements or deformations of images in practical applications. In addition, CNN can automatically extract the essential features of input data. By increasing the number of convolutional layers and pooling layers, the network can learn more complex feature expressions, thereby improving the expressive power and generalization ability of the model. CNNs have strong generalization capabilities and can demonstrate good classification performance on unseen data, thanks to their ability to extract more representative features through multi-layer convolution and pooling operations. Finally, CNNs have excellent spatial perception and can automatically learn local features of images or other grid-structured data, which is crucial for tasks such as image matching and fusion. Through these advantages, the present invention can effectively extract local feature data from visible light and infrared images, providing a solid foundation for subsequent image fusion and analysis, thereby improving the accuracy and efficiency of mine gas leakage detection.

[0050] S300 , extracting image feature points from the local feature data and performing multimodal image matching to obtain matching spatial position information data between the visible light image and the infrared image.

[0051] It should be noted that multimodal image matching refers to the process of establishing correspondence between overlapping areas of images due to differences in scale, rotation, affine and radiation between remote sensing images of different sensor types, different imaging times or viewpoints. This process is crucial for photogrammetry and remote sensing applications such as image registration, image fusion, and change detection. The challenge of multimodal image matching lies in dealing with significant geometric deformations and nonlinear grayscale differences between images of different modalities. The present invention particularly relates to feature point matching in multimodal image matching technology. Feature point matching is an important part of multimodal image matching, which involves extracting key feature points from images and finding matching point pairs between different images. Feature point matching can be based on methods such as graph matching and point set registration, which find the optimal solution by optimizing the overall correspondence matrix.

[0052] In this embodiment, the purpose of extracting image feature points from local feature data and performing multimodal image matching is to find the corresponding relationship between the visible light image and the infrared image, that is, feature point matching. These feature points can be corner points, edge points, or other significant image structures. After the matching process is completed, we obtain the matching relationship between the corresponding feature points in the two images. These matching relationships contain spatial position information data and are the basis for subsequent image fusion work. They provide accurate spatial correspondence for subsequent feature fusion.

[0053] Preferably, image feature points are extracted from local features by using the LoFTR local feature matching algorithm.

[0054] It should be noted that the LoFTR local feature matching algorithm is an advanced feature matching technology that can directly extract feature points from images and establish correspondences by utilizing the global receptive field and self-attention mechanism of the Transformer architecture without relying on traditional feature detectors. The advantage of this method in the present invention is that it can generate dense feature point matches and maintain high robustness even in areas with low texture or repeated texture, while simplifying the feature matching process because it eliminates the independent steps of feature description and matching. LoFTR can also learn globally consistent matching priors so that the transformed feature representation is relevant to the context and position, thereby improving the matching accuracy. In addition, LoFTR refines the matching items at the sub-pixel level through a correlation refinement method, further improving the matching accuracy. Therefore, LoFTR provides a powerful feature point matching tool for the mine gas leakage detection technology in the present invention, which can achieve efficient and accurate multimodal image matching in complex environments, which is crucial to improving the efficiency and accuracy of mine safety monitoring.

[0055] In a preferred embodiment, the specific steps of extracting image feature points from local features through the LoFTR local feature matching algorithm include: S210, the local feature CNN extracts a coarse feature map (e.g., 1 / 8 of the image size) and a fine feature map (e.g., 1 / 2 of the image size) from the image pair. These feature maps contain local feature information of the image, where the coarse feature map provides more global information and the fine feature map provides more detailed information; S220, the coarse feature map is flattened into a one-dimensional vector and position encoding is added. These vectors are then processed by the LoFTR module, which includes self-attention and cross-attention layers. This step makes the feature representation relevant to the context and position, enhancing the global consistency of the features; S230, through the differentiable matching layer, LoFTR can match the transformed features and generate a confidence matrix Pc. This matrix contains the confidence scores of all possible matches. Based on the confidence threshold and mutual neighbor condition, matching items are selected from Pc to generate rough matching predictions Mc; S240, for each selected rough prediction, local windows are cropped from the fine-level feature map, and the matching is refined to the sub-pixel level in these local windows as the final matching predictions Mf.

[0056] S400 , based on the spatial position information data, extracting respective deep feature data from the visible light image and the infrared image, wherein the deep feature data includes deep semantic information of the image.

[0057] In this embodiment, the purpose of extracting respective deep feature data from the visible light image and the infrared image is to capture the deep semantic information of the image and use this information to perform effective image analysis and processing.

[0058] It should be noted that deep semantic information refers to the high-level, abstract features contained in an image. These features include not only basic low-level features such as pixel values, color, and texture, but also more complex patterns and structures formed by combining these low-level features, such as object shape, outline, and semantic content. These deep features are crucial for tasks such as understanding image content, image classification, and object detection and recognition.

[0059] Preferably, based on the spatial position information data, respective deep feature data are extracted from the visible light image and the infrared image through a convolutional neural network.

[0060] It is understandable that convolutional neural networks (CNNs) are able to automatically learn feature representations from simple to complex through their multi-layered structures. At lower levels, CNNs may learn primary features such as edges, corners, and basic textures; while at higher levels, they are able to recognize more complex shapes, object parts, and even semantic information of entire objects. This hierarchical feature extraction capability enables CNNs to perform well in image recognition and classification tasks because it can capture the essential features of the image, thereby improving the accuracy and robustness of recognition. In the present invention, by extracting these deep features, safety hazards in mines, such as gas leaks, can be more accurately analyzed and identified, thereby achieving efficient monitoring and timely response to mining areas.

[0061] S500: Convert deep feature data from the spatial domain to the frequency domain using a fast Fourier transform algorithm to obtain complex feature data.

[0062] In this embodiment, "complex feature data" refers to the features obtained after converting the image from the spatial domain to the frequency domain through the fast Fourier transform. These features are in complex form because the result of the Fourier transform includes amplitude and phase information, which together constitute the real part and imaginary part of the complex number. In the present invention, the deep feature data is converted from the spatial domain to the frequency domain through the fast Fourier transform algorithm to obtain complex feature data, so as to better fuse information from different modalities. Specifically, in the process of converting the deep feature data from the spatial domain to the frequency domain through the fast Fourier transform algorithm, each feature point in the spatial domain is converted into a complex representation in the frequency domain, which contains the frequency component, amplitude and phase information of the feature point.

[0063] It should be noted that the Fast Fourier Transform (FFT) is an efficient algorithm for calculating the Discrete Fourier Transform (DFT) and its inverse transform. It decomposes large-scale DFT calculations into multiple small-scale DFT calculations through recursion and divide-and-conquer methods, significantly reducing the amount of computation and lowering the complexity from O(N^2) to O(NlogN). In the present invention, deep feature data is converted from the spatial domain to the frequency domain via the FFT. This process has multiple advantages: First, the FFT greatly improves computational efficiency, making frequency domain analysis of image features possible, especially when processing large-scale data. Second, the FFT can reveal the frequency components of an image, which is very useful for analyzing image texture, edges, and other periodic patterns. In addition, the application of the FFT in spectrum analysis can help extract the spectral features of an image, providing important data support for subsequent image processing and analysis tasks, such as image fusion and feature enhancement. The FFT can also extract the spectrum of a signal, which is very important in the fields of spectrum analysis and signal processing. Finally, the application of FFT can improve the accuracy of mine safety monitoring. By analyzing the spectral characteristics of microseismic signals, it can separate multi-source signal components, providing reliable data and analysis methods for the classification and identification of mine seismic signals and the establishment of a feature sample library. In summary, the application of FFT in this invention not only improves the efficiency of data processing, but also enhances the accuracy and real-time performance of mine gas leakage detection, providing strong technical support for mine safety.

[0064] S600: Weighting the complex feature data through a Fourier multilayer perceptron.

[0065] It should be noted that the Fourier Multilayer Perceptron (FMLP) is a neural network structure that combines Fourier transform and multilayer perceptron. It maps the input data to a higher-dimensional space through Fourier feature mapping, so that the network can learn high-frequency functions in the data. In the present invention, the advantage of weighting complex feature data through Fourier multilayer perceptron is that FMLP can process complex data in the frequency domain, and by adjusting the parameters of the Fourier feature mapping, the frequency range of neural network learning can be controlled, thereby improving the network's ability to express image features. This method makes FMLP more efficient when processing data with periodic patterns because it operates directly in the frequency domain and utilizes the sensitivity of Fourier transform to periodic signals. In addition, FMLP enhances the network's ability to learn high-frequency details through Fourier feature mapping, which is particularly important for capturing subtle features and texture information in images, thereby improving the accuracy of image fusion and analysis. In this way, FMLP not only improves the processing efficiency of deep feature data, but also enhances the model's ability to understand and analyze complex image content, which is a key technological advancement for the mine gas leakage detection technology in this invention.

[0066] S700 , converting the weighted complex feature data from the frequency domain back to the spatial domain through an inverse fast Fourier transform algorithm.

[0067] Step S700 includes the following specific steps:

[0068] S710 , converting the weighted complex feature data from the frequency domain back to the spatial domain through an inverse fast Fourier transform algorithm to obtain fused time domain feature data.

[0069] It should be noted that the Inverse Fast Fourier Transform (IFFT) algorithm is the inverse process of the Fast Fourier Transform (FFT), converting frequency-domain signals back to time-domain signals. In the present invention, the purpose of converting weighted complex feature data from the frequency domain back to the spatial domain using the IFFT algorithm is to restore the original spatial information of the image for further image processing and analysis. This conversion is an indispensable step in image processing, particularly in frequency-domain filtering, image enhancement, and feature extraction. The application of the IFFT enables complex image processing operations performed in the frequency domain to be converted back to the spatial domain, thereby obtaining an improved image. Using the IFFT, the frequency components of an image can be adjusted in the frequency domain, such as to enhance image detail or remove noise, and then the restored image can be obtained through an inverse transform. Furthermore, the IFFT is also used in image compression technology. For example, the JPEG image compression standard uses the DCT (closely related to the FFT) to convert image data from the spatial domain to the frequency domain, followed by quantization and encoding to achieve compression. Therefore, the advantage of IFFT in the present invention is that it can accurately restore the image data processed in the frequency domain to the spatial domain, providing clear image data for subsequent image reconstruction and analysis, which is crucial to improving the accuracy and efficiency of mine safety monitoring.

[0070] S720 . Perform image reconstruction using the fused time-domain feature data to obtain a fused image. The fused image includes semantic information data of the visible light image and temperature information data of the infrared image.

[0071] S730: Post-process the fused image to improve the quality of the fused image.

[0072] It is understandable that post-processing of the fused image, including contrast adjustment, color correction, etc., can improve the visual effect and application performance of the fused image.

[0073] S800: Construct a mine gas leakage detection model, and train the mine gas leakage detection model using the data obtained in step S700.

[0074] In this embodiment, the model is trained by adjusting the weights and biases of the model using the data obtained in step S700 through optimization techniques such as backpropagation and gradient descent to minimize the loss function. During the training process, hyperparameters such as the learning rate, batch size, and network architecture may need to be adjusted to improve model performance.

[0075] S900: Perform model verification on the trained mine gas leakage detection model, and adjust the trained mine gas leakage detection model according to the verification result to obtain a detection model that can be used to detect mine gas leakage.

[0076] Preferably, the trained mine gas leakage detection model is validated by a cross-validation method.

[0077] Preferably, a test set is constructed using the data obtained in step S700, and the model is validated using the test set to evaluate the model's performance, such as accuracy, recall, etc. Based on the validation results, the model structure or hyperparameters may need to be further adjusted to improve the model's generalization ability.

[0078] In one application scenario, by deploying the detection model obtained in step S900 for detecting gas leakage in mines into an actual application environment, it can be used to monitor safety hazards in mining areas, such as gas leakage, in real time.

[0079] Compared with the prior art, the present invention has the following beneficial effects:

[0080] The present invention provides a method for detecting gas leaks in mines based on Fourier transforms. First, through multimodal image acquisition, combined with visible light and infrared imaging technologies, the semantic information of visible light images and the temperature information of infrared images are fully utilized to enhance the detection capability of leakage areas. Second, multimodal image matching is performed using LoFTR to accurately find the correspondence between the two images, laying a solid foundation for subsequent feature extraction and fusion. Next, multidimensional analysis of image features is achieved through deep features extracted by the CNN neural network and frequency domain features converted by FFT, while weighted processing by the FMLP neural network further enhances the cross-domain feature fusion effect. The inverse Fourier transform converts the fused frequency domain information back to the time domain, providing rich data support for image reconstruction and fusion. On this basis, through model training, verification, and optimization, an intelligent model capable of accurately identifying and analyzing safety hazards in mining areas is constructed. Finally, the trained model is deployed in practical applications, achieving real-time monitoring and analysis of mining areas, effectively preventing and reducing the occurrence of mine safety accidents.

[0081] Overall, the Fourier transform-based mine gas leakage detection method described in the present invention not only improves the accuracy and real-time performance of mine gas detection, but also enhances the generalization ability and adaptability of the model through the application of deep learning technology, providing strong technical support for mine safety management, and has important practical application value and broad market prospects.

[0082] like Figure 3 As shown, the present invention also provides a mine gas leakage detection device based on Fourier transform, which is configured to execute the mine gas leakage detection method based on Fourier transform. The mine gas leakage detection device based on Fourier transform includes:

[0083] (1) Image acquisition module.

[0084] The image acquisition module is used to obtain visible light images and infrared images of the mine to be inspected.

[0085] (2) Local feature extraction module.

[0086] The local feature extraction module is used to extract local feature data from the visible light image and the infrared image respectively. The local feature data includes coarse-grained features and fine-grained features.

[0087] (3) Multimodal image matching module.

[0088] The multimodal image matching module is used to extract image feature points from local feature data and perform multimodal image matching to obtain spatial position information data of the matching between the visible light image and the infrared image.

[0089] (4) Deep feature extraction module.

[0090] The deep feature extraction module is used to extract respective deep feature data from the visible light image and the infrared image based on the spatial position information data. The deep feature data includes the deep semantic information of the image.

[0091] (5) Feature fusion module.

[0092] The feature fusion module is used to convert deep feature data from the spatial domain to the frequency domain through the fast Fourier transform algorithm to obtain complex feature data; weight the complex feature data through the Fourier multilayer perceptron; and convert the weighted complex feature data from the frequency domain back to the spatial domain through the inverse fast Fourier transform algorithm.

[0093] (6) Model construction and training module.

[0094] The model building and training module is used to build a mine gas leakage detection model, and train the mine gas leakage detection model using the data obtained in step S700.

[0095] (7) Model verification module.

[0096] The model verification module is used to perform model verification on the trained mine gas leakage detection model, and adjust the trained mine gas leakage detection model according to the verification result to obtain a detection model that can be used to detect mine gas leakage.

[0097] Figure 4A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0098] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0099] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0100] Processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 executes the various methods and processes described above, such as solving a Fourier transform-based method for detecting gas leaks in mines.

[0101] In some embodiments, a method for detecting gas leakage in a mine based on Fourier transform may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for detecting gas leakage in a mine based on Fourier transform described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute a method for detecting gas leakage in a mine based on Fourier transform in any other appropriate manner (e.g., by means of firmware).

[0102] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0103] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0104] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0105] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0106] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0107] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0108] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements a mine gas leakage detection method based on Fourier transform as provided in the embodiment of the present invention.

[0109] The computer program product may be implemented by writing computer program code for performing the operations of the present invention in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0110] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0111] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A mine gas leakage detection method based on Fourier transform, characterized in that: The method comprises: S100, obtaining a visible light image and an infrared image of the mine to be inspected; S200, extracting local feature data from the visible light image and the infrared image, respectively, wherein the local feature data includes coarse-grained features and fine-grained features; S300, extracting image feature points from the local feature data and performing multimodal image matching to obtain matching spatial position information data between the visible light image and the infrared image; S400, extracting respective deep feature data from the visible light image and the infrared image based on the spatial position information data, wherein the deep feature data includes deep semantic information of the image; S500, converting the deep feature data from the spatial domain to the frequency domain using a fast Fourier transform algorithm to obtain complex feature data; S600, weighting the complex feature data through a Fourier multilayer perceptron; S700, converting the weighted complex feature data from the frequency domain back to the spatial domain through an inverse fast Fourier transform algorithm; S800: Constructing a mine gas leakage detection model, and training the mine gas leakage detection model using the data obtained in step S700; S900 , performing model verification on the trained mine gas leakage detection model, and adjusting the trained mine gas leakage detection model according to the verification result to obtain a detection model for detecting mine gas leakage.

2. The method for detecting gas leakage in a mine based on Fourier transform according to claim 1, characterized in that: In step S200, local feature data of each is extracted from the visible light image and the infrared image respectively through a convolutional neural network.

3. The method for detecting gas leakage in a mine based on Fourier transform according to claim 1, characterized in that: In step S300, image feature points are extracted from the local features using the LoFTR local feature matching algorithm.

4. The method for detecting gas leakage in a mine based on Fourier transform according to claim 1, characterized in that: In step S400, based on the spatial position information data, respective deep feature data are extracted from the visible light image and the infrared image through a convolutional neural network.

5. The method for detecting mine gas leakage based on Fourier transform according to claim 1, characterized in that: Step S700 includes the following specific steps: S710, converting the weighted complex feature data from the frequency domain back to the spatial domain using an inverse fast Fourier transform algorithm to obtain fused time domain feature data; S720 . Perform image reconstruction using the fused time-domain feature data to obtain a fused image, where the fused image includes the semantic information data of the visible light image and the temperature information data of the infrared image.

6. The method for detecting gas leakage in mines based on Fourier transform according to claim 5, characterized in that: Step S700 further includes: S730: Post-process the fused image to improve the quality of the fused image.

7. The method for detecting gas leakage in a mine based on Fourier transform according to claim 1, characterized in that: In step S900, the trained mine gas leakage detection model is validated by a cross-validation method.

8. A mine gas leakage detection device based on Fourier transform, characterized in that: The mine gas leakage detection device based on Fourier transform is configured to execute a mine gas leakage detection method based on Fourier transform, and the mine gas leakage detection device based on Fourier transform includes: An image acquisition module, which is used to acquire visible light images and infrared images of the mine to be inspected; A local feature extraction module, configured to extract local feature data from the visible light image and the infrared image, respectively, wherein the local feature data includes coarse-grained features and fine-grained features; A multimodal image matching module, configured to extract image feature points from the local feature data and perform multimodal image matching to obtain spatial position information data of the matching between the visible light image and the infrared image; A deep feature extraction module, configured to extract respective deep feature data from the visible light image and the infrared image based on the spatial position information data, wherein the deep feature data includes deep semantic information of the image; A feature fusion module is configured to convert the deep feature data from the spatial domain to the frequency domain using a fast Fourier transform algorithm to obtain complex feature data; weight the complex feature data using a Fourier multilayer perceptron; and convert the weighted complex feature data from the frequency domain back to the spatial domain using an inverse fast Fourier transform algorithm; A model building and training module, which is used to build a mine gas leakage detection model and train the mine gas leakage detection model using the data obtained in step S700; The model verification module is used to perform model verification on the trained mine gas leakage detection model, and adjust the trained mine gas leakage detection model according to the verification result to obtain a detection model for detecting mine gas leakage.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, and the computer program is The at least one processor executes so that the at least one processor can execute the mine gas leakage detection method based on Fourier transform as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer readable storage medium stores a computer program, The computer program is used to enable a processor to implement a mine gas leakage detection method based on Fourier transform as described in any one of claims 1 to 7 when executed.

Images