A method for identifying and detecting illegal construction in the subway protection area based on image recognition

Abstract

The present invention discloses a method for identifying and detecting illegal construction in the subway protection area based on image recognition, which specifically relates to the technical field of image recognition. By collecting and analyzing the image data of the construction scene in the subway protection area, detecting and classifying the target equipment, extracting the spectral characteristics of the camouflage materials and evaluating the degree of limited perception range, and combining with the fluctuation of the equipment construction plan data, it accurately judges whether there is a camouflage behavior of the equipment. Once a camouflage behavior is identified, the system prompts the construction personnel to stop the operation by linking the acoustic-optic alarm device, uploads the characteristics of the camouflage materials to the database, dynamically optimizes the spectral perception model of the camera, and improves its detection ability for new camouflage behaviors, effectively solving the problem of monitoring failure caused by camouflage materials and perception blind spots, greatly improving the accuracy of illegal construction detection in the subway protection area and the safety and reliability of the system, so as to ensure the stable operation of subway facilities.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and particularly relates to a method for identifying and detecting illegal construction in a subway protection area based on image recognition. Background Art

[0002] The identification and detection of illegal construction in a subway protection area based on image recognition refers to using image recognition technology to monitor and judge possible illegal construction behaviors within the subway protection area, and realizing all-weather and high-precision automatic supervision through the method of electronic sentinels. The subway protection area usually refers to a certain area around the subway line, and construction activities within this area may pose a threat to the safety and stability of subway facilities. As an intelligent monitoring node, the electronic sentinel can automatically detect unauthorized or non-compliant construction activities by collecting and analyzing on-site images or videos in real time, and actively issue warnings or initiate corresponding protection measures. For example, at a construction site within the subway protection area, intelligent electronic sentinel cameras are installed. These cameras can not only capture high-definition images all day long, but also have intelligent analysis and edge computing capabilities. They will transmit data to the image recognition system in real time, and identify specific construction behaviors such as excavation equipment and hoisting operations through algorithms. If the electronic sentinel detects an unapproved excavator operation in a certain area, it will immediately mark it as abnormal, automatically send an alarm signal to the management department, and record the behavior time and specific location at the same time, which is convenient for quick positioning and handling. This combination of electronic sentinel technology and image recognition algorithms not only greatly reduces the cost of manual inspections, but also significantly improves the efficiency and accuracy of construction supervision within the protection area, effectively reducing the operation risk of subway facilities.

[0003] The existing technology has the following deficiencies:

[0004] The construction party may cover the surface of the equipment with special materials, making it only reflect light of a specific wavelength (such as infrared or ultraviolet), thereby effectively avoiding the detection of intelligent electronic sentinel cameras. These materials can reduce the characteristic reflection of the equipment in the visible light band, or change its appearance texture, making it impossible for the image recognition algorithm to effectively extract key features. For example, an excavator may look like an ordinary object, or become blurred in the surveillance video. In addition, if the spectral perception range of the intelligent electronic sentinel camera is limited and cannot detect the spectral characteristic changes caused by the camouflage material, the illegal construction equipment may be "invisible" in the monitoring system, making it impossible for the system to locate and identify the camouflage behavior of the key equipment. This will directly lead to the failure to detect illegal construction behaviors within the subway protection area in a timely manner, seriously weakening the reliability and security of the monitoring system, and may ultimately lead to major safety hazards such as damage to subway facilities and interruption of operation. Summary of the Invention

[0005] The object of the present invention is to provide a method for identifying and detecting illegal construction in the subway protection area based on image recognition to solve the deficiencies in the background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: A method for identifying and detecting illegal construction in the subway protection area based on image recognition, comprising the following steps:

[0007] S1: Deploy intelligent electronic sentry cameras in the subway protection area, continuously collect image data of the construction scene through the cameras, and record the timestamp and geographical location information of the images during collection;

[0008] S2: Preprocess the collected image data, transmit the preprocessed images to the image recognition system, analyze the preprocessed image data based on deep learning algorithms, detect the target construction equipment in the construction scene and classify it;

[0009] S3: Combine the multi-spectral response ability of the intelligent electronic sentry camera with the spectral reflection characteristics of the camouflage material to analyze the spectral characteristics of the target construction equipment and evaluate the degree of limitation of the spectral perception range of the intelligent electronic sentry camera;

[0010] S4: Determine whether there is a camouflage behavior of the target construction equipment according to the degree of limitation of the spectral perception range of the intelligent electronic sentry camera and the fluctuation of the construction plan data of the target construction equipment;

[0011] S5: For the identified camouflage behavior, the management platform links the on-site sound and light alarm device to remind the construction personnel to stop the operation immediately, and upload the spectral characteristics of the identified camouflage material to the spectral reflection characteristic database to optimize the camera spectral perception model.

[0012] Preferably, in S3, after analyzing the multi-spectral response ability of the intelligent electronic sentry camera, a camera spectral response anomaly index is generated. The acquisition method of the camera spectral response anomaly index is as follows:

[0013] Collect the spectral response data matrix X of the camera, where each row represents the spectral response curve of different cameras at each wavelength; standardize each column of the matrix X: ; is the standardized data, is the mean of the j-th column, is the standard deviation of the j-th column, represents the element in the matrix X, located at the position of the i-th row and the j-th column, calculate the covariance matrix C of the standardized data matrix Z: ; In the formula, T is the matrix transpose, is an n×n matrix, representing the covariance between each band; perform eigenvalue decomposition on the covariance matrix C to obtain the eigenvalues and the corresponding eigenvector ; The decomposition expression is: ; The eigenvalue represents the variance contribution degree of the k-th principal component, and the eigenvector represents the direction of the k-th principal component;

[0014] Project the standardized data Z onto the first p principal components: ; Y is the representation of the data in the principal component space, with size m×p, is the matrix containing the first p eigenvectors, with size n×p; Calculate the camera spectral response anomaly index of each sample in the principal component space, and the expression is: ; In the formula, DFC is the camera spectral response anomaly index, is the principal component vector of the i-th sample, with size 1×p, is the mean vector of all samples in the principal component space, with size 1×p, and S is the covariance matrix in the principal component space, with size p×p.

[0015] Preferably, in S3, after analyzing the spectral reflection characteristics of the camouflage material, a spectral reflectance deviation index is generated. The method for obtaining the spectral reflectance deviation index is:

[0016] Let the spectral reflectance curve of the camouflage material be , where represents the reflectance at the w-th wavelength; Let the reference standard spectral reflectance curve be , where represents the reflectance at the m-th wavelength;

[0017] Define the local distance of the spectral reflectance at a certain wavelength point: ; is the local distance between the spectral reflectance curve Q of the camouflage material at the a-th wavelength point and the reference spectral reflectance curve R at the b-th wavelength point. Construct a cumulative distance matrix D of size w×m, where each element D(a,b) represents the minimum cumulative distance from the starting point to the current point. The calculation formula is: ; The boundary condition is ; v is an index used to represent the accumulation process in the cumulative distance calculation. Starting from the lower right corner D(w,m) of the matrix, trace back along the path with the minimum cumulative distance until reaching the starting point D(1,1): The path is recorded as , where L is the path length; Calculate the spectral reflectance deviation value according to the cumulative distance D(w,m) of the cumulative path P, and the expression is: ; In the formula, EDH is the spectral reflectance deviation index.

[0018] Preferably, in S3, the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is evaluated as follows: the abnormal index of the camera spectral response and the spectral reflectance deviation index are converted into a comprehensive feature vector, and the comprehensive feature vector is used as the input of the machine learning model. The machine learning model takes the prediction of the score value label of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera for each group of comprehensive feature vectors as the prediction target, and takes minimizing the sum of the prediction errors of the score value labels of the degree of limitation of the spectral perception range of all intelligent electronic sentry cameras as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence and then the model training is stopped. The score value of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is determined according to the model output result, where the machine learning model is a polynomial regression model.

[0019] Preferably, the score value of the degree of limitation of the spectral perception range of the obtained intelligent electronic sentry camera is compared with the gradient standard threshold. The gradient standard threshold includes a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold. The score value of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is respectively compared with the first standard threshold and the second standard threshold;

[0020] If the score value of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is greater than the second standard threshold, it indicates that the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is high, and a first-level warning signal is generated at this time;

[0021] If the score value of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it indicates that the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is medium, and a second-level warning signal is generated at this time;

[0022] If the score value of the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is less than the first standard threshold, it indicates that the degree of limitation of the spectral perception range of the intelligent electronic sentry camera is low, and a third-level warning signal is generated at this time.

[0023] Preferably, in S4, after analyzing the time deviation between the actual construction behavior of the target construction equipment and the predetermined construction plan, a construction time deviation value is generated. The method for obtaining the construction time deviation value is as follows:

[0024] Obtain the actual construction time series, expressed as a vector , where is the start or end time point of the construction equipment in the q-th time period;

[0025] Obtain the predetermined construction plan time series, expressed as a vector , where is the start or end time point of the q-th planned time period, and calculate the Euclidean distance between the actual construction time vector and the scheduled construction plan time vector: ; where D is the construction time deviation value within the q-th planned time period; if the construction behavior is divided into multiple task segments, the deviation needs to be calculated separately for each task segment and accumulated: ; where is the construction time deviation value, g is the total number of task segments, is the number of time points of the h-th task segment.

[0026] Preferably, the restricted degree score value Lw of the spectral sensing range of the intelligent electronic sentry camera and the construction time deviation value are used as the input items of fuzzy logic, and whether there is a camouflage behavior of the target construction equipment is used as the output item. The judgment results of the camouflage behavior of the target construction equipment include: no camouflage behavior, possible camouflage behavior, and definite camouflage behavior;

[0027] Define the fuzzy sets of the input and output items;

[0028] Use triangular or trapezoidal membership functions to represent the membership degrees of the input and output fuzzy sets;

[0029] Establish a fuzzy rule base according to the relationship between the input items and the output items;

[0030] The output result of each rule is weighted according to the activation intensity;

[0031] Convert the fuzzy result into a definite output value;

[0032] According to the result of defuzzification , judge whether there is a camouflage behavior of the target construction equipment:

[0033] : no camouflage behavior; : possible camouflage behavior; : definite camouflage behavior.

[0034] In the above technical solution, the technical effects and advantages provided by the present invention:

[0035] 1. The present invention realizes the comprehensive monitoring and precise analysis of construction equipment in the subway protection area by combining intelligent electronic sentry cameras, multi-spectral response capabilities, deep learning algorithms, and fuzzy logic. By collecting image data through the camera and detecting and classifying the target construction equipment, and combining the evaluation of the limited degree of spectral perception range and the analysis of the fluctuation of construction plan data, it can quickly judge whether there is a camouflage behavior of the construction equipment. Once a camouflage behavior is identified, the system generates a hierarchical warning signal, and timely reminds the construction personnel to stop the operation through an audible and visual alarm device, and at the same time uploads the spectral characteristics of the camouflage material to dynamically optimize the detection model, thereby improving the system's detection ability for new camouflage behaviors.

[0036] 2. The present invention effectively solves the problems of monitoring blind spots and limited spectral perception caused by camouflage materials in the prior art, significantly improves the accuracy and efficiency of detecting illegal construction behaviors in the subway protection area, and reduces the safety risks of subway facilities. At the same time, by adopting an analysis method combining machine learning and fuzzy logic, the system has high adaptability and flexibility, can cope with complex and changeable construction environments, and provides intelligent and scalable technical support for the safety management of the subway protection area. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0038] Figure 1 It is a flowchart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0040] Embodiment, please refer to Figure 1 As shown, a method for identifying and detecting illegal construction in the subway protection area based on image recognition in this embodiment includes the following steps:

[0041] S1: Deploy intelligent electronic sentry cameras in the subway protection area, continuously collect image data of the construction scene through the cameras, and record the time stamp and geographical location information of the images during collection;

[0042] S2: Preprocess the collected image data, transmit the preprocessed images to the image recognition system, analyze the preprocessed image data based on deep learning algorithms, detect the target construction equipment in the construction scene and classify it;

[0043] S3: Combine the multi-spectral response ability of the intelligent electronic sentry camera with the spectral reflection characteristics of the camouflage material to analyze the spectral characteristics of the target construction equipment and evaluate the degree of limitation of the spectral perception range of the intelligent electronic sentry camera;

[0044] S4: According to the degree of limitation of the spectral perception range of the intelligent electronic sentry camera and the fluctuation of the construction plan data of the target construction equipment, determine whether there is a camouflage behavior of the target construction equipment;

[0045] S5: For the identified camouflage behavior, the management platform links the on-site sound and light alarm device to remind the construction personnel to stop the operation immediately, and upload the spectral characteristics of the identified camouflage material to the spectral reflection characteristic database to optimize the camera spectral perception model.

[0046] In S1, according to the scope of the subway protection area, the distribution characteristics of construction activities and risk assessment, reasonably arrange intelligent electronic sentry cameras to ensure coverage of high-risk areas (such as above subway tunnels, around pipelines, etc.). Determine the camera layout density so that the monitoring areas of adjacent cameras overlap seamlessly to avoid monitoring blind spots.

[0047] The function configuration of the intelligent camera includes: High-definition imaging ability: Equipped with a high-definition lens, supporting high-resolution (such as above 4K) image and video acquisition to ensure clear details. Multi-spectral perception ability: Integrated visible light, infrared and ultraviolet spectral perception modules to identify equipment or behaviors that may use camouflage materials. Dynamic zoom function: Having automatic zoom and perspective adjustment functions, capable of real-time tracking of dynamic construction targets. Edge computing ability: Integrated edge computing module, which can process part of the image data locally to reduce transmission latency.

[0048] The camera continuously collects real-time image data of the construction scene, supports round-the-clock monitoring, including night infrared imaging and stable acquisition under bad weather conditions. The built-in clock module automatically adds accurate timestamps to each frame of the image. The built-in GPS module or through area calibration automatically records the geographical location information of the image acquisition point to ensure the accuracy of event tracking and positioning.

[0049] The collected data is transmitted in real time to the background management platform via a 5G network or a dedicated wireless network. An encrypted transmission protocol (such as TLS / SSL) is used to protect data security and prevent data from being tampered with or intercepted during transmission. In the event of a network interruption, the camera can locally cache the data and automatically upload it after the network is restored. The camera is built with a motion detection module that automatically activates image collection and marks relevant events when it detects that construction equipment has entered the protected area. In combination with an early warning mechanism, higher-frequency image data can be preferentially collected when the target equipment enters a sensitive area.

[0050] In this application, through the above layout and functional configuration, the intelligent electronic sentry camera can efficiently collect construction scene data within the subway protected area and mark it with timestamp and geographical location information to achieve precise positioning and real-time monitoring of construction behaviors. This design provides basic data support for subsequent image analysis and abnormal behavior detection.

[0051] S2: Preprocess the collected image data, transmit the preprocessed images to the image recognition system, and analyze the preprocessed image data based on deep learning algorithms to detect and classify target construction equipment in the construction scene.

[0052] Preprocess the collected image data to improve image quality, optimize the data format for input into the model, and meet the requirements of complex construction scenarios. Specifically, this includes: removing random noise in the image through median filtering, mean filtering, or adaptive filtering algorithms, especially in images collected in low-light or dusty environments. Optimizing images with insufficient lighting conditions using histogram equalization or brightness enhancement algorithms to ensure that equipment edges and details are clearly visible. Enhancing the saliency of construction equipment in complex backgrounds through contrast stretching or adaptive contrast enhancement techniques. Uniformly adjusting the image to the input size of the deep learning model (such as 416×416 or 224×224 pixels) to ensure the efficiency and consistency of model calculations. Converting the image from RGB format to grayscale or other color spaces (such as HSV or LAB) according to model requirements to highlight equipment features. Generating diverse data samples by slightly rotating, mirror-flipping, and randomly cropping the image to enhance the robustness of the model. Introducing actual noise in the simulated scene (such as blurring, interference lines, etc.) into the image to improve the model's performance in the real environment.

[0053] Transmit the preprocessed images to the image recognition system to ensure efficient and secure data transmission. Losslessly compress the image data to reduce the transmission bandwidth occupancy while maintaining image quality. Use a high-speed communication network (such as 5G or fiber optic) to achieve low-latency real-time transmission. Protect the integrity and confidentiality of the image data during transmission through an encryption protocol (such as AES or RSA).

[0054] Analyze the preprocessed image data based on deep learning algorithms to complete the detection and classification of target construction equipment, including: Use object detection models (such as YOLO, Faster R-CNN, SSD, etc.) to detect equipment in the construction scene. Input: Preprocessed images. Output: Detection boxes (Bounding Box) and the position coordinates of the target equipment. Detect multiple devices simultaneously in the same scene to ensure comprehensive identification of equipment in complex construction site environments. Through multi-frame continuous detection combined with optical flow algorithms, achieve the tracking of the movement trajectories of equipment, providing support for subsequent behavior analysis.

[0055] Use classification models (such as ResNet, Inception, EfficientNet, etc.) to classify the detected target equipment. Classification categories: Excavation equipment (such as excavators, drills). Lifting equipment (such as cranes, tower cranes). Transportation equipment (such as trucks, tracked vehicles). Use a deep feature extraction module (such as the output of the convolutional layer) to extract the key features of the target equipment and compare them with the authorized equipment features in the construction plan database.

[0056] Output the category, position, and size information of the detected equipment. Attach a confidence score to each recognition result to indicate the confidence value of the model for this classification. Overlay the recognition results (such as annotation boxes and category labels) on the original image to facilitate managers' intuitive understanding of the recognition results.

[0057] Store the detection and classification results in the system database as input data for subsequent behavior analysis and camouflage detection. If an unauthorized device is detected, trigger the system alarm mechanism and send the preprocessed image and analysis results to the management platform.

[0058] For example, through the combined use of the YOLOv5 object detection model, optical flow algorithm, and ResNet50 classification model, achieve real-time detection, classification, and movement trajectory tracking of excavators, trucks, and cranes in the construction scene, specifically:

[0059] Input data: Video frames collected by the camera, after preprocessing (denoising, enhancement, normalization), are input into the object detection model.

[0060] Single-frame instance image: The image contains an excavator, a truck, and a crane, with a complex construction scene as the background.

[0061] Use YOLOv5 (You Only Look Once) as the object detection model.

[0062] The model extracts features and performs convolutional processing on the image to generate detection results.

[0063] Output the detection bounding box (Bounding Box) of each target device and its position coordinates in the image: Excavator: Bounding Box = [50, 100, 200, 300]. Truck: Bounding Box = [220, 120, 350, 300]. Crane: Bounding Box = [400, 50, 550, 250]. The model successfully detected three devices in the image and generated the detection bounding box and position coordinates for each device.

[0064] Combine the object detection and optical flow algorithm (Optical Flow). Conduct correlation analysis on the detection results of consecutive multiple frames, and track the position changes of the devices through the optical flow algorithm.

[0065] Output the movement trajectories of each device: Excavator: Moved from [50, 100] to [55, 105]. Truck: Moved from [220, 120] to [230, 130]. Crane: Fixed position, did not move. The movement trajectories show that the excavator and the truck are moving, while the crane remains stationary, which conforms to the actual situation of the construction site.

[0066] Use ResNet50 (Residual Network) as the classification model. For each detected target device, extract the image sub-region from the detection bounding box area. Use the convolutional neural network (CNN) to extract deep features and conduct classification: Excavator: Classification result = "Excavation equipment" (confidence 98%). Truck: Classification result = "Transportation equipment" (confidence 95%). Crane: Classification result = "Lifting equipment" (confidence 96%). The model accurately classifies the categories of the excavator, truck, and crane, and the confidence levels are all relatively high.

[0067] Obtain the feature records of authorized devices from the construction plan database: Authorized device categories: Excavator (device number: EX001, feature: yellow painting). Truck (device number: TR002, feature: blue painting). Crane (device number: CR003, feature: red painting). Compare the detected device features: The detected excavator matches the authorized device EX001 (painting consistent, features match). The detected truck matches the authorized device TR002 (painting consistent, features match). The detected crane does not match any authorized devices (painting is green, features are abnormal). The excavator and the truck are authorized devices, and the crane is not on the authorized list and may be an unauthorized device.

[0068] Equipment Detection and Classification Results: Excavators, trucks, and cranes were successfully detected and classified. An unauthorized device (green crane) was identified. The system alerted the on-site personnel to stop the operation of the crane through an audible and visual alarm device. The image, location, and unmatched feature information of the crane were uploaded to the management platform. The system stored the feature data of the crane (green painting, lifting equipment category) in the database to optimize the model for detecting future camouflage behavior.

[0069] S3: Combine the multi-spectral response ability of the intelligent electronic sentry camera with the spectral reflection characteristics of the camouflage material to analyze the spectral characteristics of the target construction equipment and evaluate the degree of limitation of the spectral perception range of the intelligent electronic sentry camera.

[0070] Generate monochromatic light within different wavelength ranges (e.g., 300 - 1100 nm, including ultraviolet, visible light, and near-infrared) through a standard light source (such as a spectrometer). Test the imaging effect and signal intensity of the intelligent electronic sentry camera at each wavelength, and record the spectral response curve. Obtain the sensitivity curve of the camera in the ultraviolet, visible light, and near-infrared ranges, indicating the wavelength range and blind area of the perception ability. Collect images under different spectral conditions in the actual construction scenario to test the spectral perception effect, such as the night infrared monitoring ability and the imaging clarity in the daytime visible light environment. Evaluation indicators include: Multi-spectral coverage range: Whether the wavelength range that the camera can effectively perceive meets the detection requirements. Pixel accuracy: The resolution and quality of imaging in the edge wavelength range.

[0071] After analyzing the multi-spectral response ability of the intelligent electronic sentry camera, generate the camera spectral response anomaly index. The method for obtaining the camera spectral response anomaly index is as follows:

[0072] Collect the spectral response data matrix X of the camera, where each row represents the spectral response curve of different cameras at each wavelength, with a size of m×n, m being the number of samples (measurement data of different cameras or at different times), and n being the number of spectral bands (e.g., the number of sampling points from 300 nm to 1100 nm).

[0073] Normalize each column of the matrix X (i.e., the data of each band) so that its mean is 0 and the standard deviation is 1: ; Z is the normalized data, is the mean of the j-th column, is the standard deviation of the j-th column, represents the element in matrix X, located at the position of the i-th row and j-th column. Calculate the covariance matrix C of the normalized data matrix Z: ; In the formula, T is the matrix transpose, is an n×n matrix representing the covariance between each band; perform eigenvalue decomposition on the covariance matrix C to obtain the eigenvalues and the corresponding eigenvector ; The decomposition expression is: ; Eigenvalue represents the variance contribution degree of the k-th principal component, and the eigenvector represents the direction of the k-th principal component;

[0074] Project the standardized data Z onto the first p principal components (usually select the number of principal components that makes the cumulative variance contribution rate reach 90% - 95%): ; Y is the representation of the data in the principal component space, with size m×p, is the matrix containing the first p eigenvectors, with size n×p;

[0075] Calculate the camera spectral response anomaly index of each sample in the principal component space, and the expression is: ; In the formula, DFC is the camera spectral response anomaly index, is the principal component vector of the i-th sample, with size 1×p, is the mean vector of all samples in the principal component space, with size 1×p, and S is the covariance matrix in the principal component space, with size p×p.

[0076] The larger the camera spectral response anomaly index, the more obvious the limitations in the spectral perception range of the intelligent electronic sentry camera. The response in some bands deviates significantly from the normal spectral characteristics. This usually means that the perception ability of the camera in these bands is severely limited. For example, it cannot effectively detect the absorption or reflection bands commonly used for camouflage materials (such as near-infrared or ultraviolet), resulting in the possible "invisibility" of camouflaged devices. This anomaly will directly reduce the recognition accuracy of the camera for construction equipment and increase the difficulty of detecting illegal construction.

[0077] On the contrary, the smaller the camera spectral response anomaly index, the more comprehensive the spectral perception range and the more consistent with the expected response, without obvious blind spots. This indicates that the camera can effectively perceive the target characteristics in most bands. Even in the face of the interference of camouflage materials, it can accurately identify through multi-spectral data. Therefore, cameras with a smaller anomaly index have higher adaptability and monitoring reliability in complex scenarios.

[0078] Collect samples of camouflage materials that the construction party may use, including infrared absorption coatings, ultraviolet reflection films, and other special coatings. Use a spectrometer to measure the reflectance curves of each camouflage material in the range of 300 - 1100 nm. Record the reflection enhancement or absorption characteristics of the camouflage materials within a specific wavelength range. Coat the surface of model equipment (such as a small excavator) with the camouflage materials and conduct multispectral imaging tests in a controlled environment. Compare the images of the camouflaged equipment captured by the intelligent electronic sentry camera with those of normal equipment to analyze the camouflage effect. Store the spectral reflection characteristics of the camouflage materials obtained from the tests in a spectral database to form a camouflage feature model for subsequent comparison.

[0079] After analyzing the spectral reflection characteristics of the camouflage materials, a spectral reflectance deviation index is generated. The method for obtaining the spectral reflectance deviation index is as follows:

[0080] Let the spectral reflectance curve of the camouflage material be , where represents the reflectance at the w-th wavelength; let the reference standard spectral reflectance curve be , where represents the reflectance at the m-th wavelength;

[0081] Define the local distance of the spectral reflectance at a certain wavelength point: ; is the local distance between the spectral reflectance curve Q of the camouflage material at the a-th wavelength point and the reference spectral reflectance curve R at the b-th wavelength point. Construct an accumulation distance matrix D of size w×m, where each element D(a,b) represents the minimum cumulative distance from the starting point to the current point. The calculation formula is: ; The boundary condition is ; v is an index used to represent the accumulation process in the cumulative distance calculation. Starting from the lower right corner D(w,m) of the matrix, trace back along the path with the minimum cumulative distance until reaching the starting point D(1,1): The path is denoted as , where L is the path length; Calculate the spectral reflectance deviation value according to the cumulative distance D(w,m) of the cumulative path P. The expression is: ; In the formula, EDH is the spectral reflectance deviation index.

[0082] The larger the spectral reflectance deviation index, the more significant the deviation between the spectral reflectance curve of the camouflage material and the reference standard. This indicates that the perception ability of the intelligent electronic sentry camera in some key spectral bands is severely limited and fails to effectively capture the true characteristics of the camouflage material. For example, the camouflage material may exhibit abnormal reflectance changes in the camera's blind spots (such as the near-infrared or ultraviolet bands), but the camera fails to respond, resulting in the monitoring system's difficulty in identifying camouflage behavior. In this case, the spectral perception range of the camera is highly restricted, threatening the reliability of the monitoring.

[0083] The smaller the spectral reflectance deviation index, the closer the spectral reflectance curve of the camouflage material is to the reference standard, and the camera can accurately capture the spectral characteristics of the camouflage material. This indicates that the intelligent electronic sentry camera has good sensing capabilities in multiple spectral bands and can detect the subtle features and changes in the reflection characteristics of the camouflage material, thus effectively avoiding the interference of camouflage behavior on the monitoring system. In this case, the spectral sensing range of the camera is less restricted, and the monitoring system is more reliable and adaptable.

[0084] Convert the spectral response anomaly index and spectral reflectance deviation index of the camera into a comprehensive feature vector, and use the comprehensive feature vector as the input of the machine learning model. The machine learning model takes the prediction of the spectral sensing range restriction degree score value label of the intelligent electronic sentry camera for each group of comprehensive feature vectors as the prediction target, and takes minimizing the sum of the prediction errors of the spectral sensing range restriction degree score value labels of all intelligent electronic sentry cameras as the training target. Train the machine learning model until the sum of the prediction errors reaches convergence and then stop the model training. Determine the spectral sensing range restriction degree score value of the intelligent electronic sentry camera according to the model output result, where the machine learning model is a polynomial regression model.

[0085] The method for obtaining the spectral sensing range restriction degree score value of the intelligent electronic sentry camera is as follows: Obtain the corresponding function expression from the comprehensive feature vector training data of the trained machine learning model: ; In the formula, is the output function of the model, DFC is the spectral response anomaly index of the camera, EDH is the spectral reflectance deviation index, is the spectral sensing range restriction degree score value of the intelligent electronic sentry camera.

[0086] Compare the obtained spectral sensing range restriction degree score value of the intelligent electronic sentry camera with the gradient standard thresholds. The gradient standard thresholds include a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold. Compare the spectral sensing range restriction degree score value of the intelligent electronic sentry camera with the first standard threshold and the second standard threshold respectively;

[0087] If the spectral sensing range restriction degree score value of the intelligent electronic sentry camera is greater than the second standard threshold, it indicates that the spectral sensing range of the intelligent electronic sentry camera is highly restricted and it may not be able to detect the camouflage behavior of the target construction equipment. At this time, generate a first-level warning signal;

[0088] If the scoring value of the restricted degree of the spectral sensing range of the intelligent electronic sentry camera is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it indicates that the restricted degree of the spectral sensing range of the intelligent electronic sentry camera is at a medium level, the sensing range of the camera is partially restricted, and some camouflage behaviors may be missed. At this time, a secondary warning signal is generated;

[0089] If the scoring value of the restricted degree of the spectral sensing range of the intelligent electronic sentry camera is less than the first standard threshold, it indicates that the restricted degree of the spectral sensing range of the intelligent electronic sentry camera is low, the sensing range is comprehensively covered, and the interference to camouflage behaviors is small. At this time, a tertiary warning signal is generated.

[0090] S4: According to the restricted degree of the spectral sensing range of the intelligent electronic sentry camera and the fluctuation of the construction plan data of the target construction equipment, determine whether there is a camouflage behavior of the target construction equipment.

[0091] The fluctuation of the construction plan data of the target construction equipment refers to the deviation between the actual construction behavior of the equipment and the predetermined construction plan, including changes in multiple dimensions such as location, time, and equipment type. For example, whether a certain equipment is constructing in an unauthorized area, whether it exceeds the predetermined construction time window, or whether the type of equipment used does not match the type recorded in the plan. These fluctuations may reflect abnormalities or potential violations in construction activities and are important bases for judging whether there is camouflage or unauthorized behavior of the equipment.

[0092] After analyzing the time deviation between the actual construction behavior of the target construction equipment and the predetermined construction plan, a construction time deviation value is generated. The method for obtaining the construction time deviation value is as follows:

[0093] Obtain the actual construction time series, expressed as a vector , where is the start or end time point of the construction equipment in the q-th time period.

[0094] Obtain the predetermined construction plan time series, expressed as a vector , where is the start or end time point of the q-th planned time period. Calculate the Euclidean distance between the actual construction time vector and the predetermined construction plan time vector: ; in the formula, D is the construction time deviation value in the q-th planned time period.

[0095] If the construction behavior is divided into multiple task segments (such as different construction areas or equipment), the deviation needs to be calculated separately for each task segment and accumulated: ; in the formula, is the construction time deviation value, g is the total number of task segments, is the number of time points in the h-th task segment.

[0096] Take the restricted degree score value Lw of the spectral perception range of the intelligent electronic sentry camera and the construction time deviation value as the input items of fuzzy logic, and take whether there is a camouflage behavior of the target construction equipment as the output item. The judgment results of the camouflage behavior of the target construction equipment include: no camouflage behavior, possible camouflage behavior, and definite camouflage behavior.

[0097] Define the fuzzy sets, including input fuzzy sets: the restricted degree score value Lw of the spectral perception range: Low, Medium, High. The construction time deviation value : Small, Medium, Large.

[0098] Output fuzzy set: the judgment result of the camouflage behavior: No disguise, Possible disguise, Definite disguise.

[0099] Use triangular or trapezoidal membership functions to represent the membership degrees of the input and output fuzzy sets.

[0100] Example: the restricted degree score value Lw of the spectral perception range: ; ; ;

[0101] The construction time deviation value Define the membership function similarly, for example:

[0102] Small: trapezoidal membership function, range [0, 10], center [0, 5].

[0103] Medium: triangular membership function, range [5, 20], peak 12.5.

[0104] Large: trapezoidal membership function, range [15, ∞], starting 20.

[0105] Establish a fuzzy rule base according to the relationship between the input items and the output items. For example:

[0106] When the spectral perception range Lw is Low and is Small, the target construction equipment is judged as having no camouflage behavior (No disguise).

[0107] When Lw is Medium and is Small, the target construction equipment may have a camouflage behavior (Possible disguise).

[0108] When Lw is High and is Small, there may be a Possible disguise behavior of the target construction equipment.

[0109] When Lw is High and is Medium, there is a Definite disguise behavior of the target construction equipment.

[0110] When Lw is High and is Large, there is a Definite disguise behavior of the target construction equipment.

[0111] According to the membership degree of the input Lw and calculate the activation strength of each rule (take the minimum value of the membership degrees of the input items).

[0112] The output result of each rule is weighted according to the activation strength.

[0113] Using the centroid method, convert the fuzzy result into a definite output value: ; where is the activation strength of rule i, is the output value corresponding to rule i (for example, 0 means no disguise, 0.5 means possible disguise, and 1 means definite disguise).

[0114] According to the defuzzified result , judge whether there is a disguise behavior of the target construction equipment:

[0115] : No disguise behavior; : There may be a disguise behavior; : There is a definite disguise behavior.

[0116] S5: For the identified disguise behavior, the management platform links the on-site audible and visual alarm devices to remind the construction personnel to stop the operation immediately, and uploads the spectral characteristics of the identified disguise materials to the spectral reflection characteristic database to optimize the spectral perception model of the camera.

[0117] When the intelligent electronic sentry camera detects a disguise behavior and judges the severity of the disguise behavior through a fuzzy logic or scoring mechanism, the management platform immediately issues an alarm signal.

[0118] The acoustic-optical alarm device includes a high-pitched horn and a high-brightness flash lamp, which respectively give the following warnings: Sound alarm: Continuously or intermittently play voice prompts, such as "Disguise behavior is detected. Please immediately stop the construction operation." Light alarm: Attract the attention of construction workers through red or yellow flash lamps to avoid safety risks caused by continuing construction due to disguise behavior. Force the on-site construction workers to stop illegal construction through strong prompt signals to ensure the safety of the subway protection area.

[0119] The management platform extracts the reflection characteristics of the disguise material from the images and spectral data collected by the camera, which usually includes the following content:

[0120] Key band range: For example, the reflection or absorption characteristics of the disguise material in the infrared, ultraviolet, or visible light band. Spectral curve characteristics: Standardize the spectral reflectance curve of the material to form characteristic data.

[0121] Upload the extracted spectral characteristic data to the spectral reflection characteristic database to store the detailed characteristics of the disguise material for subsequent detection and analysis. The data storage fields include: The band characteristics of the disguise material. Records of the time and location where the material appears. Response data during camera detection.

[0122] The database is updated dynamically. As more disguise material characteristic data is entered, the database gradually expands. Each update marks the version of the database so that the latest disguise material characteristic data can be selected when the model is called.

[0123] Combined with the disguise material characteristics uploaded to the database, update the spectral perception model of the camera to optimize the detection ability for disguise materials: Model training data enhancement: Use the new disguise material characteristics as training data to retrain the model and improve the model's recognition ability for new types of disguise materials. Update the spectral matching algorithm in the model so that it can quickly detect behaviors similar to the newly added disguise material characteristics.

[0124] If the camera has edge computing capabilities, the optimized spectral perception model is sent to the camera side through a wireless network to enhance the local detection ability of the camera. After optimization, the camera can: Improve the sensitivity to a specific spectral range (such as infrared, ultraviolet). More accurately identify the spectral abnormal characteristics of the disguise material. Iterate the spectral perception model regularly to ensure that the camera can adapt to the rapid change trend of disguise materials, such as the emergence of new coatings or materials with special reflection characteristics.

[0125] In this embodiment, in the subway protection zone, intelligent electronic sentry cameras are deployed to continuously collect image data of the construction scene, and the timestamp and geographic location information are recorded. The collected image data is pre-processed and then transmitted to the image recognition system, and the target equipment in the construction scene is detected and classified using a deep learning algorithm. Combining the multi-spectral response capability of the camera and the spectral reflectance characteristics of the camouflage material, the spectral characteristics of the target equipment are analyzed to evaluate the degree of limitation of the camera's spectral perception range. Based on the degree of limitation of the camera's perception range and the fluctuation of the construction plan data of the target equipment, it is determined whether there is camouflage behavior. If camouflage behavior is detected, the management platform will link the on-site sound and light alarm device to remind the construction personnel to stop the operation, and upload the spectral characteristics of the camouflage material to the database to optimize the camera's spectral perception model and improve the detection capability and reliability of the system.

[0126] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A method for identifying and detecting illegal construction in the subway protection area based on image recognition, characterized in that: It includes the following steps: S1: Deploy intelligent electronic sentry cameras within the subway protection area, continuously collect image data of the construction scene through the cameras, and record the timestamp and geographical location information of the images during collection; S2: Preprocess the collected image data, transmit the preprocessed images to the image recognition system, analyze the preprocessed image data based on deep learning algorithms, detect the target construction equipment in the construction scene and classify it; S3: Combine the multi-spectral response ability of the intelligent electronic sentry camera with the spectral reflection characteristics of the camouflage material, analyze the spectral characteristics of the target construction equipment, and evaluate the degree of limitation of the spectral perception range of the intelligent electronic sentry camera; Specifically, it includes: generating a camera spectral response anomaly index after analyzing the multi-spectral response ability of the intelligent electronic sentry camera, and generating a spectral reflectance deviation index after analyzing the spectral reflection characteristics of the camouflage material; converting the camera spectral response anomaly index and the spectral reflectance deviation index into a comprehensive feature vector, using the comprehensive feature vector as the input of the machine learning model, and determining the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera according to the model output result; S4: According to the degree of limitation of the spectral perception range of the intelligent electronic sentry camera and the fluctuation of the construction plan data of the target construction equipment, use the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera and the construction time deviation value as the input items of fuzzy logic; judge whether there is a camouflage behavior of the target construction equipment; S5: For the identified camouflage behavior, the management platform links the on-site sound and light alarm device to remind the construction personnel to stop the operation immediately, and upload the spectral characteristics of the identified camouflage material to the spectral reflection characteristic database to optimize the camera spectral perception model.

2. The method for identifying and detecting illegal construction in the subway protection area based on image recognition according to claim 1, wherein: In S3, the method for obtaining the camera spectral response anomaly index is: Collect the spectral response data matrix X of the camera, where each row represents the spectral response curve of different cameras at each wavelength; standardize each column of the matrix X: ; is the standardized data, is the mean of the j-th column, is the standard deviation of the j-th column, represents the element in the matrix X, located at the position of the i-th row and the j-th column, and calculate the covariance matrix C of the standardized data matrix Z: ; In the formula, T is the matrix transpose, is an n×n matrix, representing the covariance between each band; perform eigenvalue decomposition on the covariance matrix C to obtain the eigenvalues and the corresponding eigenvectors ; The decomposed expression is: ; Eigenvalue represents the variance contribution of the k-th principal component, and the eigenvector represents the direction of the k-th principal component; Project the standardized data Z onto the first p principal components: ; Y is the representation of the data in the principal component space, with size c×p, is the matrix containing the first p eigenvectors, with size n×p; Calculate the camera spectral response anomaly index of each sample in the principal component space, and the expression is: ; where DFC is the camera spectral response anomaly index, is the principal component vector of the i-th sample, with a size of 1×p, is the mean vector of all samples in the principal component space, with a size of 1×p, and S is the covariance matrix in the principal component space, with a size of p×p.

3. The method for identifying and detecting illegal construction in the subway protection area based on image recognition according to claim 2, wherein: In S3, the method for obtaining the spectral reflectance deviation index is: Let the spectral reflectance curve of the camouflage material be , where represents the reflectance at the w-th wavelength; Let the reference standard spectral reflectance curve be , where represents the reflectance at the m-th wavelength; Define the local distance of spectral reflectance at a certain wavelength point: ; Let \(d(a,b)\) be the local distance between the spectral reflectance curve \(Q\) of the camouflage material at the \(a\)-th wavelength point and the reference spectral reflectance curve \(R\) at the \(b\)-th wavelength point. Construct an accumulated distance matrix \(D\) of size \(w\times m\), where each element \(D(a,b)\) represents the minimum accumulated distance from the starting point to the current point. The calculation formula is: ; The boundary conditions are ; \(v\) is an index used to represent the accumulation process in the calculation of the accumulated distance. Starting from the lower right corner \(D(w,m)\) of the matrix, trace back along the path with the minimum accumulated distance until reaching the starting point \(D(1,1)\): The path is denoted as , where \(L\) is the path length; Calculate the spectral reflectance deviation value according to the accumulated distance \(D(w,m)\) of the accumulated path \(P\). The expression is: ; In the formula, \(EDH\) is the spectral reflectance deviation index.

4. The method for identifying and detecting illegal construction in the subway protection area based on image recognition according to claim 3, characterized in that: Compare the obtained scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera with the gradient standard thresholds. The gradient standard thresholds include a first standard threshold and a second standard threshold, and the first standard threshold is less than the second standard threshold. Compare the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera with the first standard threshold and the second standard threshold respectively; If the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera is greater than the second standard threshold, it indicates that the spectral perception range limitation degree of the intelligent electronic sentry camera is high, and at this time, a first-level warning signal is generated; If the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it indicates that the spectral perception range limitation degree of the intelligent electronic sentry camera is medium, and at this time, a second-level warning signal is generated; If the scoring value of the spectral perception range limitation degree of the intelligent electronic sentry camera is less than the first standard threshold, it indicates that the spectral perception range limitation degree of the intelligent electronic sentry camera is low, and at this time, a third-level warning signal is generated.

5. The method for identifying and detecting illegal construction in the subway protection area based on image recognition according to claim 1, wherein: In S4, after analyzing the time deviation between the actual construction behavior of the target construction equipment and the predetermined construction plan, a construction time deviation value is generated. The method for obtaining the construction time deviation value is as follows: Obtain the actual construction time series, expressed as a vector , where is the start or end time point of the construction equipment in the q-th time period; Obtain a time series of a predetermined construction plan, represented as a vector , where is the start or end time point of the q-th planned time period, and calculate the Euclidean distance between the actual construction time vector and the predetermined construction plan time vector: ; where D is the construction time deviation value within the q-th planned time period; if the construction behavior is divided into multiple task segments, the deviation needs to be calculated separately for each task segment and accumulated: ; where is the construction time deviation value, g is the total number of task segments, is the number of time points of the h-th task segment.

6. The method for identifying and detecting illegal construction in the subway protection area based on image recognition according to claim 5, characterized in that: The restricted degree score value Lw of the spectral sensing range of the intelligent electronic sentry camera and the construction time deviation value are used as input items of fuzzy logic, and whether there is a camouflage behavior of the target construction equipment is used as an output item. The judgment results of the camouflage behavior of the target construction equipment include: no camouflage behavior, possible camouflage behavior, and definite camouflage behavior; Define the fuzzy sets of the input items and output items; Use triangular or trapezoidal membership functions to represent the membership degrees of the input and output fuzzy sets; Establish a fuzzy rule base according to the relationship between the input items and output items; Assign weights to the output results of each rule according to the activation intensity; Convert the fuzzy results into definite output values; According to the result of defuzzification , determine whether there is a camouflage behavior of the target construction equipment: : No camouflage behavior; : There may be camouflage behavior; : There is clearly camouflage behavior.

Images