Intelligent detection system and method for foreign matters in rail transit section

Abstract

The invention discloses a system and a method for intelligently detecting foreign matters in a rail transit interval. The system comprises a data acquisition unit, a data transmission unit, an image processing and analyzing unit, a foreign matter algorithm unit, an alarm processing unit, a data recording unit and a user interface unit, the data acquisition unit is installed on a track line and is used for carrying out continuous image acquisition on a traffic interval and detecting abnormal conditions on a track in real time. The data transmission unit is used for transmitting the collected information; the image processing and analyzing unit is used for processing and analyzing the acquired data and identifying possible foreign matters; the foreign matter algorithm unit is used for identifying and positioning possible foreign matters; when the alarm processing unit detects the foreign matter, an alarm is given to related personnel in time; the data recording unit is used for recording all foreign matter detection data, the user interface unit is used for feeding back the obtained foreign matter information, monitoring and detection operation can be automatically carried out, manual intervention is not needed, and therefore efficiency is improved, and labor cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit, and in particular to an intelligent foreign body detection system and method for a rail transit section. Background Art

[0002] Rail transit refers to a type of transportation vehicle or transportation system in which operating vehicles need to travel on specific tracks. With the diversified development of train and railway technology, rail transit has shown more and more types, which are not only spread across long-distance land transportation, but also widely used in medium and short-distance urban public transportation.

[0003] With the acceleration of urbanization, many cities have begun to build various light railway systems serving the city. Therefore, rail transit has been widely used. However, when rail transit is in operation, there will be potential foreign objects on the corresponding rail transit lines. Traditional detection methods usually rely on manual inspection, which has many disadvantages:

[0004] First, traditional detection methods rely on manual inspection, which requires a lot of manpower and is labor-intensive, which can easily cause fatigue and misjudgment of workers.

[0005] Secondly, manual inspection takes a long time to complete, especially for large-scale rail transit sections, the work cycle is long and the efficiency is relatively low;

[0006] Furthermore, manual detection is easily affected by subjective factors, such as fatigue and lack of concentration, which may lead to missed detection or misjudgment, thus reducing detection accuracy.

[0007] Finally, the data generated by traditional detection methods need to be manually sorted and analyzed, which is time-consuming and labor-intensive, prone to data errors and omissions, and reduces data processing efficiency.

[0008] It can be seen that how to effectively improve the efficiency and accuracy of foreign object detection on rail transit lines is an urgent problem to be solved in this field. Summary of the invention

[0009] In view of the problems existing in the existing foreign object detection technology for rail transit lines, the purpose of the present invention is to provide an intelligent foreign object detection solution for rail transit sections. This solution can automatically detect foreign objects on rail transit lines, effectively improve the efficiency and accuracy of foreign object detection on rail transit lines, and can well solve the problems existing in the existing technology.

[0010] In order to achieve the above-mentioned object, the present invention provides a rail transit section foreign body intelligent detection system, which specifically includes a data acquisition unit, a data transmission unit, an image processing and analysis unit, a foreign body algorithm unit and an alarm processing unit;

[0011] The data acquisition unit is installed on the rail transit line and is configured to continuously collect status data of the rail transit line section and detect abnormal conditions on the track in real time;

[0012] The data transmission unit is configured to perform data interaction with the data acquisition unit and is capable of transmitting the image data acquired by the data acquisition unit;

[0013] The image processing and analysis unit is configured to perform data interaction with the data transmission unit, and is capable of processing and analyzing the acquired data to identify foreign objects that may exist in the rail transit section;

[0014] The foreign object algorithm unit is configured to perform data interaction with the image processing and analysis unit, and can identify and locate foreign objects that may exist on the track according to the results of processing and analysis by the image processing and analysis unit;

[0015] The alarm processing unit is configured to perform data exchange with the foreign object algorithm unit, and can form corresponding alarm information according to the result calculated by the foreign object algorithm unit.

[0016] In some embodiments of the present invention, the data acquisition unit includes a visual sensor module, an infrared sensor module, and a laser radar sensor;

[0017] The visual sensor module is configured to collect all-round data on the scene and obtain real-time images;

[0018] The infrared sensor module is configured to detect abnormal temperature on the track surface and identify possible hot spots and foreign objects; the infrared sensor module is installed near the track to ensure that it can capture infrared radiation from the track surface; the infrared sensor module continuously receives infrared radiation from the track surface and converts it into temperature readings; the system compares the real-time temperature readings with the preset temperature threshold. If the temperature reading exceeds the threshold, it is determined to be a temperature abnormality;

[0019] The lidar sensor is configured to accurately locate objects present on the track.

[0020] In some embodiments of the present invention, the image processing and analysis unit comprises:

[0021] An image preprocessing module, which performs denoising, grayscale conversion and edge detection on the collected image;

[0022] A feature extraction module, wherein the feature extraction module is configured to extract features from the image preprocessed by the image preprocessing module;

[0023] The image classification module is configured to classify and identify the image features extracted by the feature extraction module.

[0024] In some embodiments of the present invention, the foreign object algorithm unit is configured to accurately analyze and judge the abnormal areas on the image that are processed by the image processing and analysis unit and classified and identified. When a potential foreign object is identified, the foreign object algorithm unit transmits the information to the alarm processing unit.

[0025] In some embodiments of the present invention, the intelligent detection system further includes a data recording unit, which is configured to perform data interaction with the foreign object algorithm unit and the alarm processing unit, and is capable of recording all foreign object detection data and performing statistics and analysis.

[0026] In some embodiments of the present invention, the alarm processing unit includes an alarm trigger module and an alarm notification module; the alarm trigger module is configured to trigger an alarm mechanism and emit an alarm sound; the alarm notification module is configured to send alarm information to the outside and transmit data to a data recording module.

[0027] In some embodiments of the present invention, the intelligent detection system further includes a user interface unit, and the user interface unit is configured to be able to present the acquired foreign body information as feedback through human-computer interaction.

[0028] In order to achieve the above object, the present invention also provides a method for intelligent detection of foreign objects in a rail transit section, comprising:

[0029] S1: Acquire the status data of the on-site rail transit line section, including the real-time image data of the on-site rail transit line section, the on-site track surface temperature, and the positioning data of objects on the track;

[0030] S2: Preprocess the acquired on-site rail transit line section status data collection, first perform denoising, grayscale conversion, and edge detection on the acquired real-time image, then extract and classify the pipe features of the image features, and then classify and identify the extracted features to determine possible abnormal areas;

[0031] S3: Conduct in-depth analysis and judgment on the classified abnormal areas in the image. The abnormal areas are marked as potential hot spots or areas that do not match the normal track surface temperature. When potential foreign objects are identified, an alarm mechanism is triggered.

[0032] In some embodiments of the present invention, the foreign object intelligent detection method further includes a step of maintaining the track according to the alarm information.

[0033] In some embodiments of the present invention, in S8, the temperature of the hot spot foreign object is greater than 50 degrees Celsius.

[0034] The intelligent foreign object detection solution for rail transit sections provided by the present invention is based on the collection of on-site implementation data by front-end sensors and then integrated with an innovative foreign object recognition algorithm to achieve accurate automatic detection. In actual use, the detection system can automatically perform monitoring and detection operations without manual intervention, thereby improving efficiency and reducing labor costs. It can also achieve fast and accurate foreign object detection, greatly shortening the detection cycle, improving detection efficiency, and reducing the time of train operation interruption.

[0035] Secondly, the scheme of the present invention adopts advanced sensor technology and algorithms, which can accurately identify foreign objects on the track, reduce the probability of missed detection and false alarm, improve detection accuracy, and detect abnormal conditions and issue alarms in time through real-time monitoring of rail transit sections, which helps to deal with problems in a timely manner and improve safety;

[0036] Furthermore, the solution of the present invention can automatically analyze and process the detection data and store the foreign matter to provide data support for operation and maintenance;

[0037] Finally, the solution of the present invention has good environmental adaptability and can operate normally under different weather conditions, such as rainy and snowy weather or at night, thereby improving detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0039] Figure 1 A schematic diagram of the overall structure of the intelligent foreign body detection system for rail transit sections provided by the present invention;

[0040] Figure 2 A schematic diagram of the structure of a data acquisition unit in the intelligent foreign body detection system for rail transit sections provided by the present invention;

[0041] Figure 3 A schematic diagram of the structure of an image processing and analysis unit in the intelligent detection system for foreign objects in a rail transit section provided by the present invention;

[0042] Figure 4 A schematic diagram of the structure of an alarm processing unit in the intelligent foreign body detection system for rail transit sections provided by the present invention;

[0043] Figure 5 This is a schematic diagram of the structure of a data recording unit in the intelligent foreign object detection system for rail transit sections provided by the present invention.

[0044] Illustration Description:

[0045] Detection system 10;

[0046] Data collection unit 100, data transmission unit 200, image processing and analysis unit 300, foreign body algorithm unit 400, alarm processing unit 500, data recording unit 600, user interface unit 700;

[0047] Visual sensor module 110 , infrared sensor module 120 , lidar sensor 130 , image preprocessing module 310 , feature extraction module 320 , image classification module 330 , alarm triggering module 510 , alarm notification module 520 , data storage module 610 , data management module 620 . DETAILED DESCRIPTION

[0048] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below with reference to specific diagrams.

[0049] The lidar sensor mainly consists of a laser transmitter, a scanning system, a receiver and an information processing unit.

[0050] Among them, the lidar sensor emits laser pulses in the direction of the track through a laser transmitter. These laser pulses have extremely high monochromaticity, brightness and resolution, and can penetrate certain obstacles and reflect back.

[0051] The scanning system is responsible for scanning the laser beam at a certain angle and speed to cover the entire track area; the receiver receives the laser signal reflected from the object on the track. After photoelectric conversion and amplification, these signals are converted into electrical signals for subsequent processing; the information processing unit analyzes and processes the received electrical signals. First, the round-trip time of the laser pulse is calculated through algorithms such as time difference measurement or phase difference measurement to determine the distance of the target object. Secondly, the angle information of the scanning system can be used to calculate the specific position of the target object in three-dimensional space.

[0052] Based on the above mechanism, the present invention provides an intelligent foreign body detection system for rail transit sections. The system can realize real-time and high-precision automatic detection of foreign bodies on rail transit sections based on the above mechanism.

[0053] See also Figure 1 , which is an example diagram of the composition principle of the intelligent foreign object detection system 10 for rail transit sections.

[0054] As can be seen from the diagram, the rail transit section foreign object intelligent detection system 10 given in this example is specifically composed of seven functional units: a data acquisition unit 100, a data transmission unit 200, an image processing and analysis unit 300, a foreign object algorithm unit 400, an alarm processing unit 500, a data recording unit 600, and a user interface unit 700.

[0055] The data acquisition unit 100 in the system is installed on the on-site rail transit line and is used to collect real-time and continuous status data of the on-site rail transit section to detect abnormal conditions on the track in real time.

[0056] The data transmission unit 200 in the system is specifically configured to perform data interaction with the data acquisition unit 100, and can transmit the data collected by the data acquisition unit 100 to other functional units in the system or other components outside the system in real time.

[0057] The image processing and analysis unit 300 in the system is configured to establish a data connection with the data transmission unit 200, perform data exchange, obtain data transmitted by the data transmission unit 200, and process and analyze the obtained data to identify foreign objects that may exist in the rail transit section.

[0058] The foreign object algorithm unit 400 in the system is configured to perform data interaction with the image processing and analysis unit 300 , and can identify and locate possible foreign objects on the track based on the results of processing and analysis by the image processing and analysis unit 300 .

[0059] The alarm processing unit 500 in the system is configured to exchange data with the foreign object algorithm unit 400, and can generate corresponding alarm information according to the result calculated by the foreign object algorithm unit. When a foreign object is detected, an alarm information is promptly sent to relevant management personnel.

[0060] The data recording unit 600 in the system is configured to perform data interaction with the alarm processing unit 500 and the foreign object algorithm unit 400, and is capable of recording all foreign object detection data and performing statistics and analysis.

[0061] The user interface unit 700 in the system is configured to perform data interaction with the data recording unit 600 and / or the alarm processing unit 500, construct corresponding human-computer interaction, and can feed back the acquired foreign object information to relevant personnel for track maintenance and management.

[0062] With respect to the above-mentioned intelligent foreign body detection system 10 for rail transit sections, the implementation process and corresponding technical features are described in detail below.

[0063] In some embodiments, the data acquisition unit 100 and the data transmission unit 200 in the system are communicatively coupled, the data transmission unit 200 and the image processing and analysis unit 300 are communicatively coupled, the image processing and analysis unit 300 and the foreign object algorithm unit 400 are communicatively coupled, the foreign object algorithm unit 400 and the alarm processing unit 500 are communicatively coupled, the alarm processing unit 500 and the data recording unit 600 are communicatively coupled, and the data recording unit 600 and the user interface unit 700 are communicatively coupled.

[0064] As a further explanation, the above-mentioned functional units are physically connected through standard hardware interfaces to ensure reliable data transmission; standard communication protocols are used for data transmission to ensure data consistency and accuracy; optimized communication protocols and data exchange formats are used to reduce data transmission delays and redundancy, thereby improving the overall processing speed of the system; the use of asynchronous communication and message queues enables each functional unit to process tasks in parallel, further improving the concurrent processing capabilities of the system.

[0065] Based on the above scheme, the image information collected by the data acquisition unit 100 is processed in sequence by the data transmission unit 200, the image processing and analysis unit 300, the foreign object algorithm unit 400, the alarm processing unit 500, and the data recording unit 600, and is finally transmitted to the user meeting unit 700.

[0066] In some embodiments, the data acquisition unit 100 in the system is preferably composed of a visual sensor module 110, an infrared sensor module 120 and a lidar sensor 130, so that various status data of the on-site rail transit line section can be acquired in real time to ensure accurate detection of foreign objects on the track.

[0067] Specifically, the visual sensor module 110 is specifically configured to collect all-round data on the scene and obtain real-time images.

[0068] As an example, the visual sensor module 110 here preferably uses a high-resolution camera. First, according to the specific terrain, line layout and monitoring requirements of the rail transit section, select a suitable location to install the high-resolution camera. Preferably, these locations should be able to cover possible foreign body intrusion areas, such as both sides of the track, intersections, and tunnel entrances and exits; considering the camera's field of view and monitoring blind spots, multi-point control is required to ensure that there are no dead angles in the entire monitoring area, and the camera is connected to the monitoring system by wire or wireless means to ensure that the data can be transmitted to the processing center in real time and stably. During the access process, it is necessary to configure network parameters, IP addresses and other information to ensure smooth communication between the camera and the monitoring system.

[0069] Then the data acquisition mode is configured to: adopt continuous acquisition mode, that is, capture and transmit image data of the monitored area in real time. This mode can ensure that the monitoring system can obtain the latest on-site conditions and promptly detect and handle potential foreign body intrusion incidents.

[0070] As a further explanation, in addition to continuous collection, you can also set a triggered collection mode according to actual needs. For example, when there are abnormal changes in light, sound, motion and other parameters in the monitoring area, the camera is triggered to collect data. This mode can reduce unnecessary data transmission and storage burdens and improve the operating efficiency of the system.

[0071] Finally, high-resolution cameras can clearly capture real-time images of the track area, including the status of key facilities such as track surfaces, turnouts, and signal machines; in addition to the track area, cameras can also monitor the surrounding environment of the rail transit area, such as tunnel walls, bridge structures, road traffic, etc. These image data help to discover potential safety hazards and abnormal conditions, such as landslides, collapses, traffic accidents, etc.; through high frame rate and high-resolution imaging capabilities, cameras can capture clear images of fast-moving targets (such as pedestrians, vehicles, etc.), which is of great value for operations such as target tracking and behavior analysis; high-resolution cameras are usually equipped with infrared night vision functions and rain and dustproof designs, and can work normally at night or in severe weather conditions. This enables the monitoring system to maintain a full range of monitoring capabilities on site under various environmental conditions.

[0072] The infrared sensor module 120 in the data acquisition unit 100 is specifically configured to detect abnormal temperature on the track surface to identify possible hot spot foreign objects.

[0073] As an example, the infrared sensor module 120 here is specifically installed at a position that can directly and effectively detect the surface temperature of the track. Preferably, these positions will be selected on the side or above the track to ensure that the sensor can receive the infrared radiation from the track surface without obstruction; the infrared sensor module 120 needs to be fixed at a predetermined position by a stable bracket or mounting structure to prevent displacement or damage caused by vibration, wind and other factors. The infrared sensor module set up in this way can continuously receive infrared radiation from the surface of the track and convert it into a temperature reading; the system compares the real-time temperature reading with the preset temperature threshold: if the temperature reading exceeds the threshold, it is determined to be a temperature abnormality.

[0074] As a further illustration, the specific implementation process of the infrared sensor module 120 detecting temperature anomalies by receiving infrared radiation emitted from the track surface is as follows:

[0075] Infrared radiation reception: The infrared detector inside the sensor receives the infrared radiation emitted by the track surface and converts it into electrical signals.

[0076] Signal processing: Amplify and filter the received electrical signals to extract temperature-related information.

[0077] Temperature calculation: Based on the relationship between the intensity of infrared radiation and temperature, the temperature of the track surface is calculated.

[0078] Abnormal judgment: The calculated temperature is compared with the preset threshold. If it exceeds the threshold range, it is judged as a temperature abnormality.

[0079] As a further explanation, in this solution, it is preferred that the infrared sensor module 120 first preliminarily determines the temperature anomaly and then further combines other sensor data (such as the image data of the visual sensor module 110) for comprehensive analysis. As an example, the corresponding process is as follows:

[0080] Preliminary judgment: After the infrared sensor module 120 detects the temperature abnormality, it transmits the abnormal information to the data processing center.

[0081] Data fusion: The data processing center fuses the temperature data from the infrared sensor with the image data from the visual sensor to obtain more comprehensive abnormality information.

[0082] Foreign object identification: Through image processing and pattern recognition algorithms, the fused data is analyzed to identify possible hot foreign objects, such as metal fragments and burning objects on the track.

[0083] As a further example, in this solution, the infrared sensor module 120 is further configured to cooperate with the visual sensor module 110 to achieve accurate identification of foreign objects in the rail transit section.

[0084] Specifically, the infrared sensor module 120 and the visual sensor module 110 can achieve the following three aspects of organic cooperation:

[0085] Data complementarity: The infrared sensor module 120 mainly detects temperature anomalies on the track surface, while the visual sensor module 110 captures image information of the track and its surrounding environment. The two provide different types of data, but complement each other and together constitute a comprehensive monitoring of the rail transit section.

[0086] Collaborative work: When the infrared sensor module 120 detects an abnormal temperature on the track surface, it can trigger the visual sensor module 110 to collect more detailed images to further confirm the abnormal situation. At the same time, the visual sensor module 110 can also provide positioning information for the infrared sensor module 120 to help quickly locate the abnormal temperature area.

[0087] Data analysis: During the data analysis phase, the data from the infrared sensor module 120 and the visual sensor module 110 can be used as inputs and fused through algorithms to improve the accuracy and efficiency of foreign object detection.

[0088] The laser radar sensor 130 in the data acquisition unit 100 is specifically configured to accurately locate foreign objects on the track to help identify the size and shape of the foreign objects.

[0089] As an example, the deployment of the laser radar sensor 130 in the track foreign object detection system is implemented by the following scheme:

[0090] Location selection: LiDAR sensors are preferably installed in locations that can fully cover the track detection area to ensure monitoring without blind spots. Deployment methods here include fixed-point installation (such as installation on elevated structures on both sides of the track) and vehicle-mounted installation (such as installation on track inspection vehicles).

[0091] Fixed installation: Ensure that the LiDAR sensor is firmly installed to avoid data errors caused by vibration or displacement during monitoring.

[0092] Calibration and debugging: Precisely calibrate the lidar sensor to ensure its accurate relative position to the track plane, and adjust parameters to adapt to the on-site environment, such as adjusting the scanning frequency, detection distance, etc.

[0093] Connection and communication: Connect the lidar sensor to the data acquisition unit, central processing system and other equipment to ensure the stability and real-time performance of data transmission.

[0094] On this basis, this solution further configures the infrared sensor module 120, the visual sensor module 110 and the laser radar sensor 130 to be linked, so that the data collected by the three modules are closely related to each other, and these data together form the basis for comprehensive monitoring of rail transit sections and foreign body detection. The corresponding specific relationships are as follows:

[0095] Data complementarity: Different types of sensors provide different types of data, but they complement each other. Infrared sensors provide temperature information, visual sensors provide image information, and lidar sensors provide distance and shape information. These data together constitute multi-dimensional monitoring of rail transit sections.

[0096] Data fusion: In the data processing stage, the data from different types of sensors can be fused to improve the accuracy and efficiency of foreign body detection. For example, the temperature anomaly information from the infrared sensor can be fused with the image information from the visual sensor to more accurately identify hot spots.

[0097] Interoperability: Different types of sensors can work together to achieve comprehensive monitoring and foreign object detection of rail transit sections. For example, when the infrared sensor detects abnormal temperature, it can trigger the visual sensor and lidar sensor for more detailed detection and positioning.

[0098] Specifically, in this solution, the laser radar sensor 130, the visual sensor module 110 and the infrared sensor module 120 are configured to cooperate with each other in a complementary and coordinated manner as follows:

[0099] LiDAR sensor 130: mainly responsible for providing accurate distance information and three-dimensional structure information, and accurately locating foreign objects on the track.

[0100] Visual sensor module 110: By capturing images, it provides rich visual information, such as color, texture, etc., which helps to identify the specific type and characteristics of foreign objects.

[0101] Infrared sensor module 120: uses infrared radiation for non-contact measurement, is sensitive to temperature, and can assist in detecting thermal anomalies or hidden heat sources on the track.

[0102] Through data fusion technology, these three sensors can achieve more comprehensive and accurate detection and identification of foreign objects on the track.

[0103] The laser radar sensor 130 determines the distance by emitting a laser beam to the target object and receiving the reflected signal, and calculating the round-trip time difference of the laser beam. In the detection of foreign objects on the track, the laser radar can:

[0104] Scan track area: Cover the entire track area with a certain scanning frequency and angle to generate 3D point cloud data.

[0105] Data processing: Perform pre-processing such as filtering and denoising on the collected point cloud data to improve data quality.

[0106] Foreign object detection: By analyzing point cloud data through algorithms, foreign objects that do not conform to the normal structure of the track are identified and accurately located.

[0107] Furthermore, the laser radar sensor 130 can accurately represent the three-dimensional shape and size of foreign objects on the track through the generated three-dimensional point cloud data. The specific implementation process includes:

[0108] Point cloud segmentation: Segment the point cloud data containing foreign objects from the entire scene.

[0109] Feature extraction: Extract the geometric features of foreign objects, such as boundaries, contours, volumes, etc.

[0110] Shape recognition: Use machine learning or deep learning algorithms to analyze the extracted features and identify the specific shape and size of foreign objects.

[0111] Even if foreign objects are small or attached to the track surface, lidar sensors can detect and identify them with their high-precision measurement capabilities.

[0112] The solution of the present invention directly uses the laser radar sensor 130 to independently complete the detection, positioning and identification of foreign objects on the track, thereby ensuring the accuracy of detection and identification and data processing efficiency. Ultimately, the detected foreign object data can be analyzed and decided by the central processing system to take corresponding countermeasures.

[0113] It should be noted here that the data collected by the laser radar sensor 130, the visual sensor module 110 and the infrared sensor module 120 in the data acquisition unit 100 are complementary and mutually verified. Subsequently, the corresponding data processing module can comprehensively utilize the advantages of various sensors through multi-sensor data fusion technology to improve the accuracy and reliability of track foreign body detection. For example, the laser radar provides accurate distance and three-dimensional structure information, the visual sensor provides rich visual feature information, and the infrared sensor provides temperature information. The combination of the three can achieve comprehensive perception and accurate identification of track foreign bodies.

[0114] like Figure 1 As shown, in some embodiments, the data transmission unit 200 in the system is used to transmit the information acquired by the visual sensor module 110, the infrared sensor module 120 and the lidar sensor 130 to the image processing and analysis unit 300 for processing.

[0115] The specific structure of the data transmission unit 200 is not limited here and can be determined according to actual needs. For example, it can be a wireless transmission unit or a wired transmission unit.

[0116] In some embodiments, the image processing and analysis unit 300 in the system is specifically configured to process the image data captured by the image acquisition module 110, and sequentially perform preprocessing (such as denoising, contrast enhancement, etc.), feature extraction (such as edge detection, texture analysis, shape recognition, etc.) and advanced image analysis on the image data captured by the image acquisition module 110 to identify specific objects or abnormal conditions in the image.

[0117] As a further supplementary explanation, for the data collected by the infrared sensor module 120 and the lidar sensor 130 in the data acquisition unit 100, the system preferably processes them through a processing unit independent of the image processing and analysis unit 300. The processing unit is parallel to the image processing and analysis unit 300 and can synchronously process the data collected by the infrared sensor module 120 and the lidar sensor 130. This can ensure both the efficiency and accuracy of data processing, and jointly provide support for subsequent decision-making.

[0118] The image processing and analysis unit 300 and the foreign object algorithm unit 400 in the system cooperate with each other to achieve more accurate foreign object recognition.

[0119] The image processing and analysis unit 300 preliminarily analyzes the image data, extracts key features, and may preliminarily determine whether there are foreign objects or abnormal conditions based on these features);

[0120] The foreign body algorithm unit 400 is configured to perform more sophisticated analysis and identification using more complex algorithms (such as machine learning models, deep learning networks, etc.) after receiving the preliminary results from the image processing and analysis unit 300. This step is intended to improve the accuracy and reliability of identification.

[0121] As a front-end processing, the image processing and analysis unit 300 can quickly respond to and screen out possible abnormal situations, providing initial clues for the foreign object algorithm unit 400. The foreign object algorithm unit 400 uses more advanced algorithms to further verify and confirm the screened results, thereby achieving more accurate foreign object identification.

[0122] As a further illustration, Figure 3 As shown, the image processing and analysis unit 300 is preferably composed of an image preprocessing module 310, a feature extraction module 320 and an image classification module 330, so as to analyze the collected images and determine the possible presence of foreign matter.

[0123] Specifically, the image preprocessing module 310 is configured to perform denoising, grayscale conversion, and edge detection on the collected image.

[0124] The image preprocessing module 310 performs denoising, grayscale conversion, and edge detection on the collected image as follows:

[0125] First of all, denoising is the first step of image preprocessing, which aims to eliminate noise in the image and improve the clarity and recognizability of the image. Noise may be introduced by various factors such as image acquisition equipment and transmission channels, which has a negative impact on image quality. There are many denoising methods. In this scheme, median filtering is used to sort the grayscale values ​​of pixels in a window and then take the middle value to replace the original pixel value. It has a good suppression effect on salt and pepper noise (impulse noise) and can better maintain the edge of the image;

[0126] Secondly, grayscale conversion is the process of converting a color image into a grayscale image. In the grayscale conversion process, the three RGB color channels of the color image are converted into one grayscale channel. After the grayscale conversion, the image changes from color to grayscale, reducing the color information, but retaining the brightness information and most of the structural information of the image. This makes the image more concise and clear, and is convenient for subsequent processing, such as edge detection, feature extraction, etc. At the same time, the processing speed of grayscale images is usually faster than that of color images because the amount of data is reduced by two-thirds;

[0127] Finally, edge detection is used to identify edge information in an image. The edge is where the grayscale changes most dramatically in an image, corresponding to the outline of an object or the dividing line between different areas. After edge detection, the edge information in the image is clearly presented, providing important feature basis for subsequent image analysis and recognition tasks. Edge detection helps to highlight important structural information in the image, such as the outline and texture of an object, thereby improving the accuracy and efficiency of subsequent processing;

[0128] As a further specific explanation, the image preprocessing module 310 performs image preprocessing as follows:

[0129] Denoising:

[0130] Implementation: Since the scene images often contain various noises (such as Gaussian noise, salt and pepper noise, etc.), denoising is the first step of preprocessing. Filters can be used for denoising, such as the median filter for removing salt and pepper noise and the Gaussian filter for removing Gaussian noise. In addition, more advanced denoising algorithms can also be considered, such as the Non-Local Means algorithm, which can effectively remove noise while maintaining image details.

[0131] Effect: The denoised image will be clearer and the noise points will be significantly reduced, providing high-quality image data for subsequent processing.

[0132] Grayscale conversion:

[0133] Implementation plan: Converting color images into grayscale images is an effective way to reduce the amount of data and calculations. Common grayscale methods include component method, maximum method, average method and weighted average method. Since the human eye is most sensitive to green, the weighted average method is usually used, that is, the weighted coefficients of the three components of R, G, and B are 0.299, 0.587, and 0.114, respectively.

[0134] Effect: Grayscale images simplify the image data, making subsequent processing more efficient while retaining the main information of the image.

[0135] Edge Detection:

[0136] Implementation: Edge detection is a key step in image processing and is used to extract edge information from images. Common edge detection operators include the Canny operator, Sobel operator, Prewitt operator, and Roberts operator. Among them, the Canny operator is widely used due to its excellent performance. The Canny edge detection algorithm includes steps such as noise suppression, gradient calculation, non-maximum suppression, and dual threshold detection.

[0137] Effect: Edge detection can clearly outline the contours of objects in the image, providing an important basis for subsequent image analysis, recognition, and segmentation.

[0138] The image processing and analysis unit can achieve the following technical effects by performing denoising, grayscale conversion and edge detection on the image through the image preprocessing module 310:

[0139] Improve image quality: Through denoising processing, reduce noise interference in the image and improve image clarity.

[0140] Reduce computational complexity: Convert color images to grayscale images to reduce the amount of data and the computational complexity of subsequent processing.

[0141] Extract key information: Through edge detection, edge information in the image can be extracted, providing important basis for subsequent image analysis, recognition and segmentation.

[0142] This image processing and analysis unit has the following technical features when implementing image preprocessing:

[0143] High efficiency: The use of efficient denoising algorithms and grayscale methods can improve processing speed while ensuring processing effects.

[0144] Accuracy: Using advanced edge detection algorithms, such as the Canny operator, it can accurately extract edge information in images.

[0145] Flexibility: The module is designed to be flexible, and the parameters of the denoising algorithm, grayscale method, and edge detection operator can be adjusted according to actual needs to meet the image preprocessing requirements in different scenarios.

[0146] Integration: Multiple processing steps such as denoising, grayscale conversion and edge detection are integrated into one module, realizing one-stop image preprocessing and improving processing efficiency and convenience.

[0147] The feature extraction module 320 in the image processing and analysis unit is configured to extract features from the image preprocessed by the image preprocessing module 310 .

[0148] Specifically, the main task of the feature extraction module 320 is to extract information useful for subsequent processing (such as classification, recognition, etc.) from the pre-processed image.

[0149] The operation process of the feature extraction module 320 is as follows:

[0150] First, color feature extraction uses methods such as color histogram, color moment, and color set to extract the color distribution information of the image. Color features are insensitive to image rotation, translation, and scale changes, but may not be able to capture local color information;

[0151] Next, texture feature extraction is performed to extract the texture information of the image through gray-level co-occurrence matrix, Tamura texture feature, autoregressive texture model, and wavelet transform method. Texture features are rotationally invariant and resistant to noise, but are sensitive to resolution changes. Further, shape feature extraction is performed to extract the shape information of the image using edge detection, contour extraction, Hough transform and other methods. Shape features can describe the geometric characteristics of objects, but are sensitive to noise and occlusion.

[0152] Finally, advanced feature extraction: such as HOG (Histogram of Oriented Gradients), LBP (Local Binary Patterns), SIFT (Scale Invariant Feature Transform) algorithms, which can extract more complex and abstract features in images and are suitable for advanced tasks such as image recognition and classification; the corresponding technical characteristics are diversity, robustness, efficiency, distinguishability, and scalability;

[0153] As an example, the corresponding implementation scheme is as follows:

[0154] Select the feature type:

[0155] Color features: such as color histogram and color moment, are used to describe the global color distribution of an image. Color features are insensitive to changes in the direction and size of the image, but may not be able to capture local features.

[0156] Texture features: Texture features are based on statistical gray-level co-occurrence matrix, energy spectrum function method, or geometry and model-based texture feature extraction methods. Texture features describe the properties of the image surface and have a certain resistance to noise, but may be affected by changes in resolution.

[0157] Shape features: including boundary-based (such as Hough transform, Fourier transform) and region-based (such as moment invariant, geometric moment features) extraction methods. Shape features can better describe the targets in the image, but are more sensitive to deformation.

[0158] Spatial relationship features: describe the spatial position or relative direction relationship between multiple targets in the image, and enhance the ability to describe the image content.

[0159] Furthermore, the feature extraction method performed by the feature extraction module 320 includes:

[0160] Linear methods: such as principal component analysis (PCA) and linear discriminant analysis (LDA), by solving the optimal orthogonal transformation, a set of new features with the largest mutual variance is obtained to achieve feature dimensionality reduction and denoising.

[0161] Nonlinear methods: such as kernel methods (KPCA, KDA), flow type learning, etc., achieve effective feature extraction in complex data distribution through nonlinear transformation and local distance measurement.

[0162] Multi-feature fusion:

[0163] Considering that a single feature may not be able to fully describe the image, the feature extraction module 320 may use multi-feature fusion technology to combine different types of features to improve the comprehensiveness and accuracy of feature representation.

[0164] The image classification module 330 in the image processing and analysis unit is configured to classify and identify the image features extracted by the feature extraction module 320 to determine possible abnormal areas.

[0165] The image classification module 330 performs the following image classification and recognition process:

[0166] 1. Feature input: The image classification module receives the output from the feature extraction module, which is the feature representation of the image. These features can be low-level features such as color, texture, shape, etc., or high-level features extracted by a deep learning model;

[0167] 2. Classifier selection: According to the task requirements and data characteristics, select a suitable classifier for image classification. Common classifiers include support vector machine (SVM), decision tree, random forest, neural network, etc. In the field of deep learning, convolutional neural network (CNN) is widely used due to its powerful processing ability for image data;

[0168] 3. Model training: Use the labeled image dataset to train the classifier. During the training process, the classifier will learn how to map the input features to the corresponding category labels. For deep learning models such as CNN, the training process usually includes steps such as forward propagation, loss calculation, back propagation, and parameter update;

[0169] 4. Classification and recognition: Input the image features to be classified into the trained classifier. The classifier will output the probability or score of the image belonging to each category. Based on these probabilities or scores, the category to which the image belongs can be determined.

[0170] As an example, the classification and recognition process of the image classification module 330 is as follows:

[0171] Feature reception and preprocessing:

[0172] First, the image classification module receives image features from the feature extraction module 320. These features may include various types of descriptors such as color, texture, shape, etc.

[0173] Perform necessary preprocessing on the received features, such as normalization and standardization, to ensure the consistency and comparability of the feature data.

[0174] Classifier selection and training:

[0175] According to the specific requirements of image classification and the characteristics of image features, select a suitable classifier. Commonly used classifiers include support vector machine (SVM), random forest, neural network (especially convolutional neural network CNN), etc.

[0176] The classifier is trained using a labeled image dataset. During the training process, the classifier learns how to map the input features to the corresponding category labels.

[0177] Feature classification and identification:

[0178] The preprocessed image features are input into the trained classifier for classification and recognition.

[0179] The classifier determines which category the image belongs to or whether it contains abnormal areas based on the feature information.

[0180] Abnormal area judgment:

[0181] If the classifier identifies the presence of abnormal regions in the image, the abnormal regions may be further located and marked.

[0182] This usually involves a more sophisticated analysis of the image, such as using a segmentation algorithm to divide the image into multiple regions and classify each region.

[0183] The image classification module 330 has the following technical features

[0184] High efficiency:

[0185] The image classification module needs to be able to quickly process large amounts of image data and provide classification and recognition results in a short period of time, which requires the classification algorithm to have high computational efficiency and optimization capabilities.

[0186] accuracy:

[0187] The accuracy of classification and recognition is the core requirement of the image classification module. The accuracy of classification and recognition can be improved by selecting a suitable classifier and optimizing the training process.

[0188] robustness:

[0189] It has a certain resistance to various changes in the image (such as lighting changes, noise interference, occlusion, etc.), ensuring accurate classification and recognition under different conditions.

[0190] Scalability:

[0191] The image classification module should be able to flexibly adapt to different classification tasks and data sets. By adjusting the classifier parameters or changing the classifier type, it can adapt to different application scenarios and needs.

[0192] Real-time:

[0193] In some application scenarios (such as security monitoring, autonomous driving, etc.), the image classification module needs to have real-time processing capabilities and be able to complete classification and recognition tasks in a very short time.

[0194] Adaptability:

[0195] As new data is continuously added and algorithms are continuously updated, the image classification module should be able to automatically adjust and optimize its classification performance to adapt to new classification requirements and challenges.

[0196] In some embodiments, the foreign object algorithm unit 400 in the system has a built-in detection algorithm, which uses the algorithm to perform in-depth analysis and judgment on the classified abnormal areas in the image. Abnormal areas refer to those areas that are recorded by the algorithm as potential hot spot foreign objects or areas that do not match the normal track surface temperature. When a potential foreign object is identified, the foreign object algorithm unit 400 transmits the information to the alarm processing unit 500.

[0197] Specifically, when the foreign object algorithm unit 400 identifies and locates a foreign object that may exist on the track, the corresponding process is as follows:

[0198] The foreign object algorithm unit 400 first receives the image data from the image processing and analysis unit 300. The image data has been pre-processed and feature extracted, and contains the visual information of the track surface.

[0199] At the same time, the foreign object algorithm unit also receives infrared sensing data and lidar detection data. Infrared sensing data provides temperature distribution information on the track surface, which helps to identify potential foreign objects that do not conform to normal temperatures. Lidar detection data provides high-precision three-dimensional spatial information, which can accurately measure the shape, size and position of objects on the track.

[0200] Furthermore, the detection algorithm built into the foreign body algorithm unit performs a comprehensive analysis of multi-source data. First, image processing technology is used to identify abnormal areas in the image, which may appear as abnormalities in color, texture or shape.

[0201] Next, combined with infrared sensing data, the temperature characteristics of the abnormal area are analyzed to determine whether there is a significant difference in temperature with the surrounding environment, so as to further confirm whether it is a potential foreign object.

[0202] Finally, the high-precision three-dimensional information provided by the lidar detection data is used to accurately locate potential foreign objects, including their position, shape, and size.

[0203] Furthermore, the foreign body algorithm unit is also configured to be able to make intelligent judgments and decisions:

[0204] The algorithm makes intelligent judgments on potential foreign objects based on preset rules and thresholds. If foreign objects are confirmed to exist, corresponding alarm information is generated.

[0205] The foreign body algorithm unit thus formed has the following technical features:

[0206] High-precision identification: Through the fusion analysis of multi-source data, the foreign object algorithm unit can identify potential foreign objects on the track with high precision, reducing false alarms and missed alarms.

[0207] Quick response: Once a foreign object is identified, the algorithm can quickly generate an alarm message to buy time for subsequent emergency processing.

[0208] Comprehensive monitoring: Combines data from multiple sensors, such as images, infrared, and lidar, to achieve comprehensive monitoring of the track surface and improve monitoring coverage and accuracy.

[0209] See also Figure 4 In some embodiments, the alarm processing unit 500 in the system is specifically composed of an alarm trigger module 510 and an alarm notification module 520, which can trigger an alarm message and quickly notify relevant personnel when the foreign object algorithm unit 400 detects a corresponding foreign object.

[0210] Specifically, the alarm triggering module 510 here is used to trigger an alarm mechanism, such as forming a corresponding sound and light alarm.

[0211] The alarm trigger module 510 first receives the foreign object detection signal from the foreign object algorithm unit 400 through a preset communication interface. These signals usually contain key information such as the type, location, and size of the foreign object.

[0212] Next, the signal analysis unit built into the alarm trigger module 510 decodes and analyzes the received signal to extract necessary alarm information.

[0213] Furthermore, the alarm triggering module 510 compares the parsed alarm information with the preset alarm conditions (such as foreign body size, position sensitivity, etc.). These alarm conditions are set according to the actual application scenario and requirements. If the alarm conditions are met, the alarm mechanism is triggered; if not, the signal is ignored or further processed (such as logging, sending a warning notification, etc.).

[0214] Finally, once it is determined that the alarm mechanism needs to be triggered, the alarm trigger module 510 will send an alarm instruction to the alarm notification module 520. At the same time, the sound and light alarm unit inside the module is activated to generate corresponding sound and light signals to attract the attention of surrounding personnel.

[0215] As a further explanation, the alarm trigger module 510 will also record relevant information of the alarm event (such as occurrence time, foreign object details, processing status, etc.) into the system log for subsequent query and analysis.

[0216] As a further illustration, the alarm trigger module 510 may also send an alarm feedback signal to other systems or devices to achieve a wider range of linkages and responses.

[0217] As a further explanation, the alarm triggering module 510 can quickly trigger the alarm mechanism when a foreign object is detected, ensuring that relevant personnel can be informed and take corresponding countermeasures in the first time.

[0218] As a further explanation, the alarm trigger module 510 cooperates with the sound and light alarm and the alarm notification module 520, so that the system can intuitively display the alarm information and send notifications to relevant personnel through various channels (such as mobile phone text messages, emails, etc.) to improve communication efficiency.

[0219] The timely alarm and response mechanism in the alarm trigger module 510 helps to reduce the impact of foreign objects on the rail system and potential safety risks, and improves overall safety and reliability.

[0220] At the same time, the alarm recording and feedback functions in the alarm trigger module 510 provide strong support for the operation management of the system, which helps managers understand the operation status and potential problems of the system and perform maintenance and optimization in a timely manner.

[0221] In conjunction with this, the alarm notification module 520 here is configured to generate corresponding alarm information based on the results detected by the foreign object algorithm unit 400 when the alarm trigger module 510 triggers the alarm mechanism, notify relevant personnel through Internet technology and transmit the data to the data recording module 600.

[0222] The alarm notification module 520 is configured to notify relevant personnel through Internet technology and transmit the data to the data recording module 600 when the alarm triggering module 510 triggers the alarm mechanism, including the following specific contents:

[0223] Alarm type: First, the data will clearly indicate the type of alarm, such as intrusion alarm, fire alarm, equipment failure alarm, etc., so that the recipient can quickly understand the nature of the alarm.

[0224] Alarm location: The data will contain the specific location information of the alarm, which may be expressed in the form of GPS coordinates, room number, device number, etc., to help the receiver quickly locate the source of the alarm.

[0225] Timestamp: records the exact time when the alarm occurs, usually in the form of year, month, day, hour, minute and second, to facilitate subsequent analysis and tracking.

[0226] Detection parameters: If the alarm is triggered by abnormalities in specific parameters detected by the foreign object algorithm unit 400, the specific values ​​and states of these abnormal parameters will also be recorded and transmitted. For example, in a fire alarm system, these may be abnormal values ​​of parameters such as smoke density and temperature.

[0227] Device status: The current status information of the relevant device, such as whether it is online or faulty, may also be included in the transmitted data.

[0228] Image or video (if applicable): In some advanced systems, if equipped with monitoring equipment such as cameras, relevant images or video clips may also be transmitted when an alarm is triggered to provide more intuitive alarm scene information.

[0229] User information: If the system is able to identify the user or relevant responsible person who triggered the alarm, their information may also be included in the transmitted data for subsequent notification and follow-up.

[0230] Other additional information: Depending on the specific design and requirements of the system, it may also include some other additional information, such as the urgency of the alarm, handling suggestions, etc.

[0231] See also Figure 5 In some embodiments, the data recording unit 600 in the system is specifically composed of a data storage module 610 and a data management module 620, thereby storing the foreign object detection data in the rail transit section and ensuring the integrity of previous data.

[0232] In the rail transit system, the data recording unit 600 is a key component, which is specifically composed of a data storage module 610 and a data management module 620. It aims to comprehensively and accurately store the foreign object detection data in the rail transit section to ensure the integrity and traceability of previous data.

[0233] As an example, the specific detection data involved in this system include:

[0234] Image Data:

[0235] Real-time video streams or key frame images from image acquisition devices such as cameras. These image data record the actual scenes within the rail transit section, including the track surface, surrounding environment, and any possible foreign objects.

[0236] Infrared sensing data:

[0237] Infrared sensors collect data on the temperature distribution of the track surface. This data can reveal areas that are out of range of normal temperatures, helping to identify potential foreign objects or anomalies.

[0238] LiDAR detection data:

[0239] The three-dimensional point cloud data generated by the LiDAR equipment provides high-precision spatial information, which can accurately depict the shape, size and position of objects on the track, and is an important basis for foreign object detection and positioning.

[0240] Alarm information:

[0241] When the foreign object algorithm unit detects a foreign object, it will generate corresponding alarm information. This alarm information includes key information such as alarm time, alarm type, foreign object location, and foreign object size, which is an important basis for subsequent emergency treatment and data analysis.

[0242] System log:

[0243] Log data that records system operation status, equipment working conditions, data processing flow, etc. These log data are important for diagnosing system failures, optimizing system performance, and ensuring data security.

[0244] Analysis results:

[0245] The data processed and analyzed by the foreign object algorithm unit, such as foreign object identification results, abnormal area marking, potential risk assessment, etc. These result data provide important decision support for rail transit operations.

[0246] The data storage module 610 in the data recording unit 600 is responsible for safely and reliably storing the above-mentioned various detection data, alarm information, system logs and analysis results in a database or storage medium. The storage method may include distributed storage, cloud storage, etc. to ensure the scalability and accessibility of data.

[0247] The data management module 620 in the data recording unit 600 is responsible for managing and maintaining the stored data, including data indexing, querying, backup, recovery and security assurance, etc. The module may also provide data visualization tools so that users can intuitively view and analyze data.

[0248] Furthermore, the data storage module 610 here is specifically configured to record and store the foreign object data collected by the visual sensor module 110 , the infrared sensor module 120 and the lidar sensor 130 .

[0249] The data storage module 610 obtains data from the visual sensor module 110, the infrared sensor module 120 and the laser radar sensor 130 through an interface or a network communication protocol. When these sensor modules detect foreign objects in the rail transit section, they send the collected data (such as images, infrared sensing data, laser radar point clouds, etc.) to the data storage module 610 according to a predetermined format and protocol.

[0250] As an example, the data storage module 610 in the data recording unit 600 completes the corresponding data recording and storage process as follows:

[0251] Data reception: The data storage module 610 first receives the data stream from the sensor module to ensure the integrity and real-time nature of the data.

[0252] Data preprocessing: Perform preliminary processing on the received data, such as denoising, format conversion, etc., for subsequent storage and analysis.

[0253] Data Storage:

[0254] Storage method: Use efficient data storage technologies, such as distributed storage systems, database management systems (DBMS), or cloud storage services to ensure data scalability and accessibility.

[0255] Storage structure: Design a reasonable storage structure based on the data type and purpose, such as relational database, non-relational database or file system, etc. For image and video data, a dedicated image database or file system may be used for storage.

[0256] Data compression: For large amounts of image and video data, data compression technology can be used to reduce storage space while maintaining data recoverability and quality.

[0257] Data index: Create data index to speed up data retrieval and improve data access efficiency.

[0258] Data backup: Back up data regularly or according to set conditions to prevent data loss or damage.

[0259] The data management module 620 in the data recording unit 600 is configured to cooperate with the data storage module 610 to manage and control the authority of data backup to ensure data security and integrity.

[0260] The data management module 620 is responsible for managing the data backup work in the data storage module 610, including:

[0261] Backup strategy formulation: Develop a reasonable backup strategy based on business needs and data importance, such as full backup, incremental backup, or differential backup.

[0262] Backup task scheduling: Automatically or manually schedule backup tasks to ensure data is backed up on time.

[0263] Backup data storage: Store backup data in secure and reliable storage media, such as tape libraries, cloud storage, etc.

[0264] Backup data verification: Regularly verify the integrity and recoverability of backup data to ensure the validity of backup data.

[0265] The data management module 620 implements authority control in the following ways:

[0266] User authentication: User identity is authenticated using methods such as username and password, and biometrics.

[0267] Permission allocation: Assign corresponding data access and operation permissions based on user roles and business needs. For example, only specific users or roles can access sensitive data or perform operations such as data backup and recovery.

[0268] Access control: Use technical means such as access control lists (ACLs) to limit user access to data and operation permissions.

[0269] The data management module 620 in the data recording unit 600 is further configured to ensure data security and integrity:

[0270] Data encryption: Encrypt sensitive data for storage and transmission to ensure data security during storage and transmission.

[0271] Access auditing: Record users’ access to and operation of data in order to track and review abnormal behavior.

[0272] Data verification: Use data verification codes (such as CRC, MD5, etc.) to verify data to ensure data integrity and consistency.

[0273] Disaster recovery and backup: Establish a disaster recovery and backup system to store backup data in multiple geographically dispersed locations to prevent data loss caused by emergencies such as natural disasters.

[0274] Security policy updates: Regularly update security policies and technical means to respond to new security threats and vulnerabilities.

[0275] In some implementations, the system may not be equipped with a corresponding data recording unit 600, depending on actual needs.

[0276] In some embodiments, the user interface unit 700 in the system is preferably presented in the form of a human-computer interaction interface, which is used to receive and display foreign object information sent by the data storage module 610, and to monitor the system's operating status, abnormal conditions, historical data and other information in real time.

[0277] The rail transit section foreign body intelligent detection system 10 thus formed is applied to rail transit. Through the cooperation between various units, the rail transit section can be detected, identified, alarmed, recorded and displayed to relevant personnel without manual inspection, which greatly improves efficiency and accuracy.

[0278] The following is a specific application example to illustrate the process of intelligent real-time detection of foreign objects in rail transit sections by the rail transit section foreign object intelligent detection system given in this example.

[0279] S1: First, the visual sensor module 110 collects all-round data on site to obtain real-time images. The infrared sensor module 120 detects abnormal temperature on the track surface and identifies possible hot spots and foreign objects. The lidar sensor 130 accurately locates objects on the track to help identify the size and shape of foreign objects.

[0280] Furthermore, in step S1, the temperature of the hot spot foreign matter is greater than 50 degrees Celsius.

[0281] Here, the temperature threshold of the hot spot foreign object is a setting for the temperature characteristics of the foreign object itself, which is used to distinguish normal objects from foreign objects whose temperature may rise due to some reason (such as friction, short circuit, etc.). In actual applications, the temperature threshold of the hot spot foreign object should be adjusted according to the specific situation to ensure that it can accurately reflect the actual state of the foreign object.

[0282] S2: The data transmission unit 200 transmits the information acquired by the visual sensor module 110, the infrared sensor module 120 and the lidar sensor 130 to the image processing and analysis unit 300, and uses the image preprocessing module 310 to perform denoising, grayscale conversion, and edge detection on the collected image, and then uses the feature extraction module 320 to extract and classify the pipe features of the image features, and further uses the image classification module 330 to classify and identify the extracted features to determine possible abnormal areas.

[0283] S3: Use the detection algorithm in the foreign object algorithm unit 400 to conduct in-depth analysis and judgment on the classified abnormal areas in the image. Abnormal areas refer to those areas marked by the algorithm as potential hot spot foreign objects or areas that do not match the normal track surface temperature. When potential foreign objects are identified, the foreign object algorithm unit 400 transmits the information to the alarm processing unit 500. At this time, the alarm trigger module 510 triggers the alarm mechanism and sounds an alarm. The alarm notification module 520 notifies relevant personnel through Internet technology.

[0284] Furthermore, in the specific implementation of this step, the foreign object algorithm unit 400 uses an image processing algorithm (such as and a machine learning model (such as a convolutional neural network CNN) to analyze the preprocessed image and identify areas that are inconsistent with the normal track surface temperature, namely, abnormal areas; these abnormal areas may appear as areas with abnormally increased (hot spots) or decreased temperatures, which are in sharp contrast to the temperature of the surrounding environment; high-quality track thermal images are obtained by image acquisition and preprocessing, and necessary preprocessing is performed; image processing algorithms and machine learning models are used to identify areas that are inconsistent with the normal track surface temperature; feature extraction and classification recognition are used to extract features from abnormal areas, and classifiers or models are used to identify and classify, and other sensor information is combined to further analyze and assess the risks of abnormal areas; an alarm mechanism is triggered according to the assessment results to notify relevant personnel to handle the situation.

[0285] Furthermore, the detection algorithm in step S3 is preferably any one of YOLO or Faster RR-CNN.

[0286] S4: Using the data storage module 610 to record and store the foreign object data collected by the visual sensor module 110, the infrared sensor module 120 and the laser radar sensor 130, and then using the data management module 620 to manage and control the backup of the data to ensure data security and integrity;

[0287] S5: Finally, the user interface unit 600 receives and displays the foreign object information. When the relevant personnel find the abnormal situation, they will maintain the track in time. The user interface unit 600 can monitor the system's operating status, abnormal situation and historical data in real time.

[0288] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent foreign body detection system for rail transit sections, characterized in that: The intelligent detection system specifically includes a data acquisition unit, a data transmission unit, an image processing and analysis unit, a foreign body algorithm unit, and an alarm processing unit; The data acquisition unit is installed on the rail transit line and is configured to continuously collect status data of the rail transit line section and detect abnormal conditions on the track in real time; The data transmission unit is configured to perform data interaction with the data acquisition unit and is capable of transmitting the image data acquired by the data acquisition unit; The image processing and analysis unit is configured to perform data interaction with the data transmission unit, and is capable of processing and analyzing the acquired data to identify foreign objects that may exist in the rail transit section; The foreign object algorithm unit is configured to perform data interaction with the image processing and analysis unit, and can identify and locate foreign objects that may exist on the track according to the results of processing and analysis by the image processing and analysis unit; The alarm processing unit is configured to perform data exchange with the foreign object algorithm unit, and can form corresponding alarm information according to the result calculated by the foreign object algorithm unit.

2. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The data acquisition unit includes a visual sensor module, an infrared sensor module and a laser radar sensor; The visual sensor module is configured to collect all-round data on the scene and obtain real-time images; The infrared sensor module is configured to detect abnormal temperature on the track surface and identify possible hot spot foreign objects; The lidar sensor is configured to accurately locate objects present on the track.

3. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The image processing and analysis unit comprises: An image preprocessing module, which performs denoising, grayscale conversion and edge detection on the collected image; A feature extraction module, wherein the feature extraction module is configured to extract features from the image preprocessed by the image preprocessing module; The image classification module is configured to classify and identify the image features extracted by the feature extraction module.

4. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The foreign object algorithm unit is configured to accurately analyze and judge the abnormal area on the image that is processed by the image processing and analysis unit and classified and identified. When a potential foreign object is identified, the foreign object algorithm unit transmits the information to the alarm processing unit.

5. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The intelligent detection system also includes a data recording unit, which is configured to interact with the foreign object algorithm unit and the alarm processing unit, and is capable of recording all foreign object detection data and performing statistics and analysis.

6. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The alarm processing unit includes an alarm trigger module and an alarm notification module; the alarm trigger module is configured to trigger an alarm mechanism and emit an alarm sound; the alarm notification module is configured to send alarm information to the outside and transmit data to a data recording module.

7. The intelligent foreign body detection system for rail transit section according to claim 1 is characterized in that: The intelligent detection system further includes a user interface unit, which is configured to provide feedback and presentation of the acquired foreign body information through human-computer interaction.

8. A method for intelligent detection of foreign objects in rail transit sections, characterized in that: include: S1: Acquire the status data of the on-site rail transit line section, including the real-time image data of the on-site rail transit line section, the on-site track surface temperature, and the positioning data of objects on the track; S2: Preprocess the acquired on-site rail transit line section status data collection, first perform denoising, grayscale conversion, and edge detection on the acquired real-time image, then extract and classify the pipe features of the image features, and then classify and identify the extracted features to determine possible abnormal areas; S3: Conduct in-depth analysis and judgment on the classified abnormal areas in the image. The abnormal areas are marked as potential hot spots or areas that do not match the normal track surface temperature. When potential foreign objects are identified, an alarm mechanism is triggered.

9. The intelligent detection method for foreign objects in rail transit sections according to claim 8 is characterized in that: The foreign object intelligent detection method also includes the step of maintaining the track according to the alarm information.

10. The intelligent foreign body detection method for rail transit section according to claim 8, characterized in that: In S1, the temperature of the hot spot foreign body is greater than 50 degrees Celsius.

Images