A method and system for detecting foreign objects in channels based on intelligent interpretation

Abstract

This invention discloses a channel foreign object monitoring method and system based on intelligent interpretation, relating to the field of image recognition technology. It solves the problem of poor monitoring effect in existing channel foreign object monitoring methods. The method includes the following steps: Step S1: Performing two-dimensional occupancy analysis on the channel monitoring area based on regional remote sensing images; screening the channel monitoring area based on the analysis results to obtain initial foreign object screening data; Step S2: Acquiring regional images of the channel re-inspection area based on the initial foreign object screening data and creating a foreign object monitoring space volume; using the foreign object monitoring space volume to monitor the channel foreign object volume in the re-inspection area, obtaining the foreign object volume occupancy corresponding to the re-inspection area, and obtaining channel foreign object re-inspection data; Step S3: Providing foreign object early warning for the channel area based on the channel foreign object re-inspection data. This invention can improve the efficiency and accuracy of channel foreign object monitoring.

Description

Technical Field

[0001] This invention belongs to the field of engineering technology and relates to image recognition technology, specifically a channel foreign object monitoring method and system based on intelligent interpretation. Background Technology

[0002] Existing technologies for detecting foreign objects in channels have the following drawbacks:

[0003] 1. Existing channel foreign object monitoring technology requires the use of cameras and millimeter-wave radar to monitor the channel area. It does not use segmented remote sensing images to perform preliminary screening of foreign objects in the channel area, and cannot perform targeted re-inspection of the channel area based on the preliminary screening results, thus resulting in low efficiency of channel foreign object monitoring methods.

[0004] 2. Existing channel foreign object monitoring technology does not acquire regional images of the channel re-inspection area and create a foreign object monitoring space, and does not use the foreign object monitoring space to monitor the volume of foreign objects in the channel re-inspection area. It is impossible to obtain the foreign object volume occupancy degree corresponding to each re-inspection area based on the monitoring results to carry out channel foreign object early warning, thus resulting in a lack of accuracy in the channel foreign object early warning results.

[0005] To address this, we propose a channel foreign object monitoring method and system based on intelligent interpretation. Summary of the Invention

[0006] In view of the shortcomings of existing technologies, the purpose of this invention is to provide a channel foreign object monitoring method and system based on intelligent interpretation, and to improve the accuracy of channel foreign object monitoring results.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: a channel foreign object detection method based on intelligent interpretation, comprising the following steps:

[0008] Step S1: Acquire remote sensing images of the channel monitoring area, perform two-dimensional encroachment analysis on the channel monitoring area based on the regional remote sensing images, and screen the channel monitoring area based on the analysis results to obtain preliminary screening data of foreign objects in the channel.

[0009] Step S2: Based on the initial screening data of foreign objects in the channel, perform regional image acquisition on the channel re-inspection area and create a foreign object monitoring space. Use the foreign object monitoring space to monitor the volume of foreign objects in the channel re-inspection area, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area, and obtain the channel foreign object re-inspection data.

[0010] Step S3: Issue a foreign object warning for the channel area based on the channel foreign object re-inspection data.

[0011] Furthermore, step S1 also includes the following steps:

[0012] Step S11: Acquire the channel area that needs to be monitored for foreign objects, obtain multiple channel monitoring areas, and acquire channel remote sensing images for each abnormal monitoring area to obtain the regional remote sensing image corresponding to each abnormal monitoring area.

[0013] Step S12: Perform image analysis on the regional remote sensing image, and obtain the channel area anomaly occupancy degree corresponding to each channel monitoring area based on the analysis results;

[0014] Step S13: Set the abnormal encroachment benchmark interval. If the abnormal encroachment degree of the channel area is within the abnormal encroachment benchmark interval, then set the corresponding channel monitoring area as the channel re-inspection area. If the abnormal encroachment degree of the channel area is not within the abnormal encroachment benchmark interval, then set the corresponding channel monitoring area as the channel without foreign objects area, and obtain the initial screening data of channel foreign objects.

[0015] Furthermore, step S12 also includes the following steps:

[0016] Step S121: Randomly select one sample monitoring area from the multiple acquired anomaly monitoring areas, and set the remote sensing image of the area corresponding to the sample monitoring area as the remote sensing image of the sample area.

[0017] Step S122: Mark the abnormal monitoring area in the remote sensing image of the sample area to obtain the marked monitoring area. Set a pixel traversal point in the marked monitoring area and obtain the pixel point at the current time to obtain the real-time monitoring pixel.

[0018] Step S123: Perform pixel depth analysis on the real-time monitored pixels, and monitor the pixel depth deviation corresponding to the pixels in real time based on the analysis results;

[0019] Step S124: Use pixel traversal points to traverse the marked monitoring area and obtain the pixel depth deviation corresponding to each pixel. If the pixel depth deviation is greater than the pixel depth deviation reference value, the pixel is divided into foreign object pixels. If the pixel depth deviation is less than or equal to the pixel depth deviation reference value, the pixel is divided into channel pixels.

[0020] Step S125: Perform clustering analysis on the foreign object pixels in the marked monitoring area to obtain the channel area abnormal occupancy degree corresponding to the marked monitoring area, and obtain the channel area abnormal occupancy degree corresponding to each abnormal monitoring area.

[0021] Furthermore, step S123 also includes the following steps:

[0022] The remote sensing image of the sample area is converted to grayscale to obtain the grayscale image of the sample area. A coordinate system is created for the marked monitoring area to obtain the grayscale region Cartesian coordinate system. The coordinates of the real-time monitoring pixels in the grayscale region Cartesian coordinate system are obtained to obtain the feature monitoring coordinates.

[0023] Historical remote sensing images are acquired for the sample monitoring area where no foreign objects are found, and the obtained historical remote sensing images are processed into grayscale to obtain multiple historical remote sensing grayscale images. A coordinate system is created in the historical remote sensing grayscale images to obtain a plane rectangular coordinate system of the historical area. The pixel points corresponding to the feature monitoring coordinates in each historical area plane rectangular coordinate system are acquired to obtain multiple historical non-object pixels, and a reasonable preset range of pixel depth is set accordingly.

[0024] The pixel depth corresponding to the real-time monitored pixel is numerically acquired to obtain the real-time monitored pixel depth. The difference between the real-time monitored pixel depth and the preset range of reasonable pixel depth is calculated to obtain the pixel depth deviation corresponding to the real-time monitored pixel.

[0025] Furthermore, step S125 also includes the following steps:

[0026] For any sample abnormal pixel in the marked monitoring area, obtain the line distance between the sample abnormal pixel and each foreign object pixel, and set the minimum line distance as the distance between the adjacent abnormal pixels corresponding to the sample abnormal pixel.

[0027] Obtain the distance between adjacent abnormal pixels corresponding to each foreign object pixel, set the abnormal pixel baseline distance, obtain foreign object pixels whose adjacent abnormal pixel distance is less than or equal to the abnormal pixel baseline distance, obtain valid foreign object pixels, and count the number of valid foreign object pixels to obtain the number of valid foreign object pixels.

[0028] The number of pixels in the marked monitoring area is counted to obtain the number of pixels in the area. The ratio of the number of effective foreign object pixels to the number of pixels in the area is calculated to obtain the abnormal occupancy degree of the channel area corresponding to the sample monitoring area.

[0029] Furthermore, step S2 also includes the following steps:

[0030] Step S21: Obtain the initial screening data of foreign objects in the channel, obtain the channel re-inspection area based on the initial screening data of foreign objects in the channel, and arbitrarily select one sample re-inspection area from the multiple channel re-inspection areas obtained.

[0031] Step S22: Perform a channel foreign object re-inspection on the sample re-inspection area, and obtain the foreign object volume occupancy degree corresponding to the sample re-inspection area based on the re-inspection results;

[0032] Step S23: Obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area to obtain channel foreign object re-inspection data;

[0033] Step S22 further includes the following steps:

[0034] Step S221: Set up a foreign object monitoring space in the spatial environment corresponding to the sample re-inspection area, create a three-dimensional coordinate system in the foreign object monitoring space, and use any one of the endpoints of the foreign object monitoring space as the origin to create a three-dimensional coordinate system, thus obtaining the foreign object monitoring three-dimensional coordinate system.

[0035] Step S222: The spatial sides contained in the foreign object monitoring space are respectively designated as the first spatial side, the second spatial side, the third spatial side, the fourth spatial side, the fifth spatial side, and the sixth spatial side;

[0036] Step S223: Image acquisition is performed on the side of the first space. Based on the acquired image, the coordinates of the effective foreign object pixels corresponding to the side of the first space are acquired in the three-dimensional coordinate system of foreign object monitoring to obtain the coordinates of the foreign object pixels corresponding to the side of the first space.

[0037] Step S224: Obtain the foreign object pixel coordinates corresponding to the second space side to the sixth space side, mark the pixel corresponding to each foreign object pixel coordinate in the foreign object monitoring space volume to obtain multiple abnormal pixel points, fill the foreign object pixel points with feature cubes, and accumulate the feature cubes to obtain the foreign object region volume value.

[0038] Step S225: Obtain the volume value of the foreign object monitoring space, calculate the ratio of the foreign object area volume value to the space volume value, and obtain the foreign object volume occupancy degree corresponding to the sample re-inspection area.

[0039] Furthermore, step S223 also includes the following steps:

[0040] Step S2231: Set up an image acquisition terminal in front of the side of the first space, and name the cameras included in the image acquisition terminal as the first terminal camera and the second terminal camera respectively.

[0041] Step S2232: Use the first terminal camera to acquire an image of the first side of the space to obtain a first side image; use the second terminal camera to acquire an image of the second side of the space to obtain a second side image.

[0042] Step S2233: Acquire the effective foreign object pixels in the first side image, select a sample foreign object pixel, and perform binocular visual analysis on the sample foreign object pixel using the first side image and the second side image to obtain the first terminal straight-line distance corresponding to the sample foreign object pixel.

[0043] Step S2234: Obtain the straight-line distance of the first terminal corresponding to each valid foreign object pixel. Obtain the straight-line distance between the first terminal camera and the side of the first space to obtain the side distance difference of the terminal. If the valid foreign object pixel is greater than or equal to the side distance difference of the terminal, then set the corresponding valid pixel as an in-plane foreign object pixel. If the valid foreign object pixel is less than the side distance difference of the terminal, then set the corresponding valid pixel as an out-of-plane foreign object pixel.

[0044] Step S2235: Obtain the coordinate position of the first terminal camera in the three-dimensional coordinate system for foreign object detection, and obtain the coordinate position of the first terminal. Based on the coordinate position of the first terminal and the straight-line distance of the first terminal, obtain the coordinate position of each foreign object pixel in the three-dimensional coordinate system for foreign object detection, and obtain the foreign object pixel coordinates corresponding to the side of the first space.

[0045] Furthermore, step S2233 also includes the following steps:

[0046] The sample foreign object pixels in the first side image are marked as the first foreign object pixel position. The sample foreign object pixels are obtained in the second side image. The first side image and the second side image are edge-aligned. The relative position of the sample foreign object pixels in the second side image is projected onto the first side image to obtain the second foreign object pixel position.

[0047] In the first side image, the straight-line distance between the first foreign object pixel position and the second foreign object pixel position is obtained to obtain the parallax of the foreign object pixel point;

[0048] The optical center distance between the first terminal camera and the second terminal camera is obtained to obtain the optical center distance of the terminal device; the focal length of the first terminal camera is obtained to obtain the focal length of the terminal device.

[0049] The straight-line distance of the first terminal is obtained by calculating the optical center distance of the terminal device, the focal length of the terminal device, and the parallax of the foreign object pixels.

[0050] Furthermore, step S3 also includes the following steps:

[0051] Obtain channel foreign object re-inspection data, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area based on the channel foreign object re-inspection data, and set a foreign object volume occupancy benchmark interval.

[0052] If the foreign object volume encroachment is within the foreign object volume encroachment benchmark range, a foreign object warning will be issued for the channel area. If the foreign object volume encroachment is not within the foreign object volume encroachment benchmark range, a foreign object warning will not be issued for the channel area.

[0053] The channel foreign object detection system based on intelligent interpretation includes:

[0054] Image acquisition module: acquires remote sensing images of the channel monitoring area, performs two-dimensional encroachment analysis on the channel monitoring area based on the regional remote sensing images, and screens the channel monitoring area based on the analysis results to obtain preliminary screening data of foreign objects in the channel;

[0055] Foreign object re-inspection module: Based on the initial screening data of foreign objects in the channel, regional images are acquired for the channel re-inspection area and a foreign object monitoring space is created. The foreign object monitoring space is used to monitor the volume of foreign objects in the channel re-inspection area, and the foreign object volume occupancy degree corresponding to each channel re-inspection area is obtained to obtain the channel foreign object re-inspection data.

[0056] Foreign object warning module: Provides foreign object warnings for the channel area based on the re-inspection data of foreign objects in the channel.

[0057] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0058] 1. This invention acquires remote sensing images of the channel monitoring area by segmenting remote sensing images, performs two-dimensional occupancy analysis of the channel monitoring area based on the regional remote sensing images, screens the channel monitoring area based on the analysis results, and conducts targeted re-inspection of the channel area based on the preliminary screening results, thereby improving the efficiency of foreign object monitoring in the channel.

[0059] 2. This invention acquires regional images of the channel re-inspection area and creates a foreign object monitoring space. The foreign object monitoring space is used to monitor the volume of foreign objects in the channel re-inspection area. Based on the monitoring results, the foreign object volume occupancy degree corresponding to the channel re-inspection area is obtained, and foreign object early warning is issued accordingly. This can effectively improve the accuracy of foreign object monitoring results. Attached Figure Description

[0060] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0061] Figure 1 This is a diagram illustrating the implementation steps of the present invention;

[0062] Figure 2 This is an overall system block diagram of the present invention;

[0063] Figure 3 This is a schematic diagram of the abnormal monitoring area of ​​the present invention;

[0064] Figure 4This is a schematic diagram of the three-dimensional coordinate system for foreign object monitoring according to the present invention. Detailed Implementation

[0065] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0066] Example 1

[0067] Please see Figure 1 This invention provides a technical solution: a channel foreign object detection method based on intelligent interpretation, comprising the following steps:

[0068] Step S1: Acquire remote sensing images of the channel monitoring area, perform two-dimensional encroachment analysis on the channel monitoring area based on the regional remote sensing images, and screen the channel monitoring area based on the analysis results to obtain preliminary screening data of foreign objects in the channel.

[0069] Step S1 further includes the following steps:

[0070] Step S11: Acquire the channel area that needs to be monitored for foreign objects, obtain multiple channel monitoring areas, and acquire channel remote sensing images for each abnormal monitoring area to obtain the regional remote sensing image corresponding to each abnormal monitoring area.

[0071] Step S12: Perform image analysis on the regional remote sensing image, and obtain the channel area anomaly occupancy degree corresponding to each channel monitoring area based on the analysis results;

[0072] Step S12 further includes the following steps:

[0073] Step S121: Randomly select one sample monitoring area from the multiple acquired anomaly monitoring areas, and set the remote sensing image of the area corresponding to the sample monitoring area as the remote sensing image of the sample area.

[0074] Step S122: Mark the abnormal monitoring area in the remote sensing image of the sample area to obtain the marked monitoring area. Set a pixel traversal point in the marked monitoring area and obtain the pixel point at the current time to obtain the real-time monitoring pixel.

[0075] Step S123: Perform pixel depth analysis on the real-time monitored pixels, and monitor the pixel depth deviation corresponding to the pixels in real time based on the analysis results;

[0076] Step S123 further includes the following steps:

[0077] The remote sensing image of the sample area is converted to grayscale to obtain the grayscale image of the sample area. A coordinate system is created for the marked monitoring area to obtain the grayscale region Cartesian coordinate system. The coordinates of the real-time monitoring pixels in the grayscale region Cartesian coordinate system are obtained to obtain the feature monitoring coordinates.

[0078] Historical remote sensing images are acquired for the sample monitoring area where no foreign objects are found, and the obtained historical remote sensing images are processed into grayscale to obtain multiple historical remote sensing grayscale images. A coordinate system is created in the historical remote sensing grayscale images to obtain a plane rectangular coordinate system of the historical area. The pixel points corresponding to the feature monitoring coordinates in each historical area plane rectangular coordinate system are acquired to obtain multiple historical non-object pixels, and a reasonable preset range of pixel depth is set accordingly.

[0079] The pixel depth corresponding to the real-time monitored pixel is numerically obtained to obtain the real-time monitored pixel depth. The difference between the real-time monitored pixel depth and the preset range of reasonable pixel depth is calculated to obtain the pixel depth deviation corresponding to the real-time monitored pixel.

[0080] Step S124: Use pixel traversal points to traverse the marked monitoring area and obtain the pixel depth deviation corresponding to each pixel. If the pixel depth deviation is greater than the pixel depth deviation reference value, the pixel is divided into foreign object pixels. If the pixel depth deviation is less than or equal to the pixel depth deviation reference value, the pixel is divided into channel pixels.

[0081] Step S125: Perform clustering analysis on the foreign object pixels in the marked monitoring area to obtain the abnormal occupancy degree of the channel area corresponding to the marked monitoring area;

[0082] Step S125 further includes the following steps:

[0083] For any sample abnormal pixel in the marked monitoring area, obtain the line distance between the sample abnormal pixel and each foreign object pixel, and set the minimum line distance as the distance between the adjacent abnormal pixels corresponding to the sample abnormal pixel.

[0084] Obtain the distance between adjacent abnormal pixels corresponding to each foreign object pixel, set the abnormal pixel baseline distance, obtain foreign object pixels whose adjacent abnormal pixel distance is less than or equal to the abnormal pixel baseline distance, obtain valid foreign object pixels, and count the number of valid foreign object pixels to obtain the number of valid foreign object pixels.

[0085] The number of pixels in the marked monitoring area is counted to obtain the number of pixels in the area. The ratio of the number of effective foreign object pixels to the number of pixels in the area is calculated to obtain the abnormal occupancy degree of the channel area corresponding to the sample monitoring area.

[0086] Step S126: Obtain the abnormal encroachment degree of the channel area corresponding to each abnormal monitoring area;

[0087] Step S13: Set the abnormal encroachment benchmark interval. If the abnormal encroachment degree of the channel area is within the abnormal encroachment benchmark interval, then set the corresponding channel monitoring area as the channel re-inspection area. If the abnormal encroachment degree of the channel area is not within the abnormal encroachment benchmark interval, then set the corresponding channel monitoring area as the channel without foreign objects area, and obtain the channel foreign object initial screening data.

[0088] It should be noted here that:

[0089] Step S1 significantly improves the intelligence and efficiency of the monitoring process. This method changes the traditional time-consuming and labor-intensive model that might require indiscriminate, high-intensity analysis of the entire area. It first uses two-dimensional images for rapid, large-scale scanning diagnosis, intelligently identifying key areas suspected of abnormal encroachment. This screening mechanism, moving from a broad "surface" to a specific "point," allows subsequent re-inspections to be targeted, concentrating limited computing resources and time on the most suspicious areas. This significantly reduces unnecessary inspection losses overall, achieving optimized channel monitoring efficiency while ensuring comprehensive coverage.

[0090] Step S2: Based on the initial screening data of foreign objects in the channel, perform regional image acquisition on the channel re-inspection area and create a foreign object monitoring space. Use the foreign object monitoring space to monitor the volume of foreign objects in the channel re-inspection area, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area, and obtain the channel foreign object re-inspection data.

[0091] Step S2 further includes the following steps:

[0092] Step S21: Obtain the initial screening data of foreign objects in the channel, obtain the channel re-inspection area based on the initial screening data of foreign objects in the channel, and arbitrarily select one sample re-inspection area from the multiple channel re-inspection areas obtained.

[0093] Step S22: Perform a channel foreign object re-inspection on the sample re-inspection area, and obtain the foreign object volume occupancy degree corresponding to the sample re-inspection area based on the re-inspection results;

[0094] Step S22 further includes the following steps:

[0095] Step S221: Set up a foreign object monitoring space in the spatial environment corresponding to the sample re-inspection area, create a three-dimensional coordinate system in the foreign object monitoring space, and use any one of the endpoints of the foreign object monitoring space as the origin to create a three-dimensional coordinate system, thus obtaining the foreign object monitoring three-dimensional coordinate system.

[0096] Step S222: The spatial sides contained in the foreign object monitoring space are respectively designated as the first spatial side, the second spatial side, the third spatial side, the fourth spatial side, the fifth spatial side, and the sixth spatial side;

[0097] Step S223: Image acquisition is performed on the side of the first space. Based on the acquired image, the coordinates of the effective foreign object pixels corresponding to the side of the first space are acquired in the three-dimensional coordinate system of foreign object monitoring to obtain the coordinates of the foreign object pixels corresponding to the side of the first space.

[0098] Step S223 further includes the following steps:

[0099] Step S2231: Set up an image acquisition terminal in front of the side of the first space, and name the cameras included in the image acquisition terminal as the first terminal camera and the second terminal camera respectively.

[0100] Step S2232: Use the first terminal camera to acquire an image of the first side of the space to obtain a first side image; use the second terminal camera to acquire an image of the second side of the space to obtain a second side image.

[0101] Step S2233: Acquire the effective foreign object pixels in the first side image, select a sample foreign object pixel, and perform binocular visual analysis on the sample foreign object pixel using the first side image and the second side image to obtain the first terminal straight-line distance corresponding to the sample foreign object pixel.

[0102] Step S2233 further includes the following steps:

[0103] The sample foreign object pixels in the first side image are marked as the first foreign object pixel position. The sample foreign object pixels are obtained in the second side image. The first side image and the second side image are edge-aligned. The relative position of the sample foreign object pixels in the second side image is projected onto the first side image to obtain the second foreign object pixel position.

[0104] In the first side image, the straight-line distance between the first foreign object pixel position and the second foreign object pixel position is obtained to obtain the parallax of the foreign object pixel point;

[0105] The optical center distance between the first terminal camera and the second terminal camera is obtained to obtain the optical center distance of the terminal device; the focal length of the first terminal camera is obtained to obtain the focal length of the terminal device.

[0106] The straight-line distance of the first terminal is obtained by calculating the optical center distance of the terminal device, the focal length of the terminal device, and the parallax of the foreign object pixels.

[0107] The straight-line distance to the first terminal is calculated using the following formula:

[0108] ;

[0109] Where Zj1 is the straight-line distance of the first terminal, Sjj is the focal length of the terminal device, Gxj is the optical center distance of the terminal device, and Zsc is the parallax of the foreign object pixel.

[0110] Step S2234: Obtain the straight-line distance of the first terminal corresponding to each valid foreign object pixel. Obtain the straight-line distance between the first terminal camera and the side of the first space to obtain the side distance difference of the terminal. If the valid foreign object pixel is greater than or equal to the side distance difference of the terminal, then set the corresponding valid pixel as an in-plane foreign object pixel. If the valid foreign object pixel is less than the side distance difference of the terminal, then set the corresponding valid pixel as an out-of-plane foreign object pixel.

[0111] Step S2235: Obtain the coordinate position of the first terminal camera in the three-dimensional coordinate system for foreign object detection, and obtain the coordinate position of the first terminal. Based on the coordinate position of the first terminal and the straight-line distance of the first terminal, obtain the coordinate position of each foreign object pixel in the three-dimensional coordinate system for foreign object detection, and obtain the foreign object pixel coordinates corresponding to the side of the first space.

[0112] Step S224: Obtain the foreign object pixel coordinates corresponding to the second space side to the sixth space side, mark the pixel corresponding to each foreign object pixel coordinate in the foreign object monitoring space volume to obtain multiple abnormal pixel points, fill the foreign object pixel points with feature cubes, and accumulate the feature cubes to obtain the foreign object region volume value.

[0113] Step S225: Obtain the volume value of the foreign object monitoring space, calculate the ratio of the foreign object area volume value to the space volume value, and obtain the foreign object volume occupancy degree corresponding to the sample re-inspection area.

[0114] Step S23: Obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area to obtain channel foreign object re-inspection data;

[0115] It should be noted here that:

[0116] Step S2 involves creating a foreign object monitoring space for volume monitoring and early warning. Its core benefit lies in significantly enhancing the accuracy of monitoring results and their decision-support value. Compared to relying solely on two-dimensional planar image analysis, which may fail to distinguish information such as shadows, materials, and height, leading to misjudgments, this method, by constructing a three-dimensional "foreign object monitoring space," can obtain the three-dimensional size information of the foreign object, thereby calculating a more physically accurate "volume occupancy." This dimensional upgrade from "planar" to "three-dimensional" allows the system to more realistically assess the actual impact of foreign objects on the channel space, effectively avoiding false alarms or missed alarms caused by missing information dimensions. The resulting early warning information is more objective and reliable, providing a more precise basis for subsequent handling decisions.

[0117] Step S3: Issue a foreign object warning for the channel area based on the channel foreign object re-inspection data;

[0118] Step S3 further includes the following steps:

[0119] Obtain channel foreign object re-inspection data, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area based on the channel foreign object re-inspection data, and set a foreign object volume occupancy benchmark interval.

[0120] If the foreign object volume encroachment is within the foreign object volume encroachment benchmark range, a foreign object warning will be issued for the channel area. If the foreign object volume encroachment is not within the foreign object volume encroachment benchmark range, a foreign object warning will not be issued for the channel area.

[0121] Example 2

[0122] Please see Figure 2 Based on another concept of the same invention, a channel foreign object monitoring method and system based on intelligent interpretation is proposed, including an image acquisition module, a foreign object re-inspection module, a foreign object early warning module and a server. The image acquisition module, the foreign object re-inspection module and the foreign object early warning module are respectively connected to the server, and the server controls the image acquisition module, the foreign object re-inspection module and the foreign object early warning module respectively.

[0123] The image acquisition module acquires remote sensing images of the channel monitoring area, performs two-dimensional encroachment analysis on the channel monitoring area based on the regional remote sensing images, and screens the channel monitoring area based on the analysis results to obtain preliminary screening data of foreign objects in the channel.

[0124] Specifically as follows:

[0125] The channel areas that need to be monitored for foreign objects are acquired, resulting in multiple channel monitoring areas. Remote sensing images of each abnormal monitoring area are then acquired to obtain the corresponding regional remote sensing image.

[0126] It should be noted here that:

[0127] In this application, the passage area referred to herein specifically refers to the road used by rail transit vehicles, including but not limited to high-speed rail tracks, train tracks, and subway tracks.

[0128] Image analysis is performed on the regional remote sensing images, and the channel area anomaly occupancy degree corresponding to each channel monitoring area is obtained based on the analysis results;

[0129] Specifically as follows:

[0130] Arbitrarily select one sample monitoring area from the multiple acquired anomaly monitoring areas, and set the remote sensing image of the area corresponding to the sample monitoring area as the sample area remote sensing image;

[0131] Please see Figure 3 The abnormal monitoring areas in the remote sensing image of the sample area are marked to obtain the marked monitoring area. A pixel traversal point is set in the marked monitoring area, and the pixel point at the current time is obtained to obtain the real-time monitoring pixel.

[0132] Perform pixel depth analysis on real-time monitored pixels, and monitor the pixel depth deviation corresponding to the pixel points in real time based on the analysis results;

[0133] Specifically as follows:

[0134] The remote sensing image of the sample area is converted to grayscale to obtain the grayscale image of the sample area. A coordinate system is created for the marked monitoring area to obtain the grayscale region Cartesian coordinate system. The coordinates of the real-time monitoring pixels in the grayscale region Cartesian coordinate system are obtained to obtain the feature monitoring coordinates.

[0135] Historical remote sensing images are acquired for the sample monitoring area where no foreign objects are found, and the obtained historical remote sensing images are processed into grayscale to obtain multiple historical remote sensing grayscale images. A coordinate system is created in the historical remote sensing grayscale images to obtain a plane rectangular coordinate system of the historical area. The pixel points corresponding to the feature monitoring coordinates in each historical area plane rectangular coordinate system are acquired to obtain multiple historical non-object pixels, and a reasonable preset range of pixel depth is set accordingly.

[0136] It should be noted here that:

[0137] In this application, the historical remote sensing images and the sample area remote sensing images involved herein have the same shooting parameters, including but not limited to image acquisition angle, image shooting focal length and image shooting distance.

[0138] In this application, the relative position of the grayscale region plane rectangular coordinate system in the sample region grayscale image is the same as the relative position of the historical region plane rectangular coordinate system in the historical remote sensing grayscale image.

[0139] The pixel depth corresponding to each historically identical pixel is obtained, resulting in multiple historical pixel depth values. The historical pixel depth value with the largest value is set as the upper limit of the reasonable pixel depth, and the historical pixel depth value with the smallest value is set as the lower limit of the reasonable pixel depth, thus obtaining a preset range of reasonable pixel depth.

[0140] The pixel depth corresponding to the real-time monitored pixel is numerically obtained to obtain the real-time monitored pixel depth. The difference between the real-time monitored pixel depth and the preset range of reasonable pixel depth is calculated to obtain the pixel depth deviation corresponding to the real-time monitored pixel.

[0141] It should be noted here that:

[0142] The pixel depth deviation corresponding to the real-time monitored pixel is obtained as follows:

[0143] If the real-time monitored pixel depth is greater than the upper limit of the reasonable pixel depth, the difference between the real-time monitored pixel depth and the upper limit of the reasonable pixel depth is calculated to obtain the pixel depth deviation corresponding to the real-time monitored pixel. If the real-time monitored pixel depth is less than the lower limit of the reasonable pixel depth, the difference between the lower limit of the reasonable pixel depth and the real-time monitored pixel depth is calculated to obtain the pixel depth deviation corresponding to the real-time monitored pixel. If the real-time monitored pixel depth is within the preset range of reasonable pixel depth, the value 0 is used to assign a parameter value to the pixel depth deviation corresponding to the real-time monitored pixel.

[0144] The pixel traversal points are used to traverse the marked monitoring area and obtain the pixel depth deviation corresponding to each pixel. If the pixel depth deviation is greater than the pixel depth deviation reference value, the pixel is divided into foreign object pixels. If the pixel depth deviation is less than or equal to the pixel depth deviation reference value, the pixel is divided into channel pixels.

[0145] It should be noted here that:

[0146] In this application, the pixel depth deviation reference value is 0.

[0147] Clustering analysis is performed on foreign object pixels in the marked monitoring area to obtain the abnormal occupancy degree of the channel area corresponding to the marked monitoring area;

[0148] Specifically as follows:

[0149] For any sample abnormal pixel in the marked monitoring area, obtain the line distance between the sample abnormal pixel and each foreign object pixel, and set the minimum line distance as the distance between the adjacent abnormal pixels corresponding to the sample abnormal pixel.

[0150] Repeat the process of obtaining the distance between adjacent abnormal pixels corresponding to the abnormal pixels in the sample, obtain the distance between adjacent abnormal pixels corresponding to each foreign object pixel, set the abnormal pixel baseline distance, obtain foreign object pixels whose distance between adjacent abnormal pixels is less than or equal to the abnormal pixel baseline distance, obtain the effective foreign object pixels, and count the number of effective foreign object pixels to obtain the number of effective foreign object pixels.

[0151] It should be noted here that:

[0152] In this application, historical remote sensing images of the marked monitoring area containing foreign objects in the channel were acquired, and the foreign objects in the channel in the historical remote sensing images were divided into multiple pixels. The distance between adjacent abnormal pixels corresponding to each pixel was obtained, and the difference between the average value and the standard deviation of the distance between adjacent abnormal pixels was calculated to obtain the baseline distance of the abnormal pixels.

[0153] The number of pixels in the marked monitoring area is counted to obtain the number of pixels in the area. The ratio of the number of effective foreign object pixels to the number of pixels in the area is calculated to obtain the abnormal occupancy degree of the channel area corresponding to the sample monitoring area.

[0154] Repeat the process of obtaining the abnormal occupancy of the channel area corresponding to the sample monitoring area, and obtain the abnormal occupancy of the channel area corresponding to each abnormal monitoring area.

[0155] Set an abnormal encroachment baseline range. If the abnormal encroachment degree of the channel area is within the abnormal encroachment baseline range, then set the corresponding channel monitoring area as the channel re-inspection area. If the abnormal encroachment degree of the channel area is not within the abnormal encroachment baseline range, then set the corresponding channel monitoring area as the channel foreign object-free area to obtain the channel foreign object initial screening data.

[0156] It should be noted here that:

[0157] In this application, the channel monitoring areas where foreign objects were found in the channel in historical monitoring were acquired, and the abnormal occupancy degree of the channel area corresponding to each channel monitoring area was acquired. The channel area abnormal occupancy degree with the smallest value was set as the lower limit of the abnormal occupancy benchmark interval, and 100% was set as the upper limit of the abnormal occupancy benchmark interval.

[0158] When the abnormal encroachment rate of the channel area is 100%, the channel monitoring area is completely occupied by foreign objects in the channel or the remote sensing acquisition device is completely blocked by abnormal objects.

[0159] The foreign object re-inspection module acquires regional images of the channel re-inspection area based on the channel foreign object initial screening data and creates a foreign object monitoring space. The foreign object monitoring space is used to monitor the channel foreign object volume in the channel re-inspection area, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area, and obtain the channel foreign object re-inspection data.

[0160] Specifically as follows:

[0161] Obtain preliminary screening data for foreign objects in the channel; based on the preliminary screening data, obtain the channel re-inspection area; and arbitrarily select one sample re-inspection area from the multiple channel re-inspection areas obtained.

[0162] Foreign matter re-inspection is performed on the sample re-inspection area, and the foreign matter volume occupancy degree corresponding to the sample re-inspection area is obtained based on the re-inspection results.

[0163] Specifically as follows:

[0164] Set up a foreign object monitoring space in the spatial environment corresponding to the sample re-inspection area, create a three-dimensional coordinate system in the foreign object monitoring space, and create a three-dimensional coordinate system by taking any one of the endpoints of the foreign object monitoring space as the origin, thus obtaining the foreign object monitoring three-dimensional coordinate system.

[0165] It should be noted here that:

[0166] In this application, the foreign object monitoring space specifically refers to the three-dimensional spatial area in the sample re-inspection area where channel foreign object monitoring is required;

[0167] In this application, the three-dimensional coordinate system for foreign object monitoring can be used to represent the coordinates of a spatial region outside the foreign object monitoring space.

[0168] The spatial sides contained in the foreign object monitoring space are respectively designated as the first spatial side, the second spatial side, the third spatial side, the fourth spatial side, the fifth spatial side, and the sixth spatial side;

[0169] Image acquisition is performed on the side of the first space. Based on the acquired image, the coordinates of the effective foreign object pixels corresponding to the side of the first space are acquired in the three-dimensional coordinate system of foreign object monitoring to obtain the coordinates of the foreign object pixels corresponding to the side of the first space.

[0170] Specifically as follows:

[0171] An image acquisition terminal is set up on the front side of the first space, and the cameras included in the image acquisition terminal are named the first terminal camera and the second terminal camera, respectively.

[0172] It should be noted here that:

[0173] In this application, the image acquisition terminal specifically refers to a binocular camera.

[0174] The first terminal camera is used to acquire images of the side of the first space to obtain a first side image, and the second terminal camera is used to acquire images of the side of the second space to obtain a second side image.

[0175] Repeat the process of acquiring effective foreign object pixels in the remote sensing image of the sample area, acquire effective foreign object pixels in the first side image, select a sample foreign object pixel, and use the first side image and the second side image to perform binocular visual analysis on the sample foreign object pixel to obtain the first terminal straight-line distance corresponding to the sample foreign object pixel.

[0176] Specifically as follows:

[0177] The sample foreign object pixels in the first side image are marked as the first foreign object pixel position. The sample foreign object pixels are obtained in the second side image. The first side image and the second side image are edge-aligned. The relative position of the sample foreign object pixels in the second side image is projected onto the first side image to obtain the second foreign object pixel position.

[0178] In the first side image, the straight-line distance between the first foreign object pixel position and the second foreign object pixel position is obtained to obtain the parallax of the foreign object pixel point;

[0179] The optical center distance between the first terminal camera and the second terminal camera is obtained to obtain the optical center distance of the terminal device; the focal length of the first terminal camera is obtained to obtain the focal length of the terminal device.

[0180] It should be noted here that:

[0181] In this application, the first terminal camera and the second terminal camera maintain the same focal length.

[0182] The straight-line distance of the first terminal is obtained by calculating the optical center distance of the terminal device, the focal length of the terminal device, and the parallax of the foreign object pixels.

[0183] The straight-line distance to the first terminal is calculated using the following formula:

[0184] ;

[0185] Where Zj1 is the straight-line distance of the first terminal, Sjj is the focal length of the terminal device, Gxj is the optical center distance of the terminal device, and Zsc is the parallax of the foreign object pixel.

[0186] Repeatedly measure the straight-line distance of the first terminal corresponding to the sample foreign object pixel, and obtain the straight-line distance of the first terminal corresponding to each valid foreign object pixel. Obtain the straight-line distance between the first terminal camera and the side of the first space to obtain the side distance difference of the terminal. If the valid foreign object pixel is greater than or equal to the side distance difference of the terminal, then the corresponding valid pixel is set as an in-plane foreign object pixel. If the valid foreign object pixel is less than the side distance difference of the terminal, then the corresponding valid pixel is set as an out-of-plane foreign object pixel.

[0187] Please see Figure 4 The coordinate position of the first terminal camera in the three-dimensional coordinate system for foreign object detection is obtained to obtain the coordinate position of the first terminal. Based on the coordinate position of the first terminal and the straight-line distance of the first terminal, the coordinate position of each foreign object pixel in the three-dimensional coordinate system for foreign object detection is obtained to obtain the foreign object pixel coordinates corresponding to the side of the first space.

[0188] It should be noted here that:

[0189] The coordinates of the foreign object pixels within the object detection 3D coordinate system are obtained based on the coordinates of the first terminal, as detailed below:

[0190] Given that the coordinates of the first terminal are (xi, yi, zi), the straight-line distance of the first terminal is d, and the first side image is in the xy plane of the three-dimensional coordinate system for foreign object monitoring;

[0191] If the in-plane foreign object pixel is located in the positive z-axis direction of the first terminal camera, the coordinate position of the in-plane foreign object pixel in the three-dimensional coordinate system of foreign object monitoring can be obtained as (xi,yi,zi+d).

[0192] If the in-plane foreign object pixel is located in the negative z-axis direction of the first terminal camera, the coordinate position of the in-plane foreign object pixel in the three-dimensional coordinate system of foreign object monitoring can be obtained as (xi,yi,zi-d).

[0193] Similarly, if the first side image is in the xz plane of the three-dimensional coordinate system for foreign object monitoring, then the y-axis coordinate is changed; if the first side image is in the yz plane of the three-dimensional coordinate system for foreign object monitoring, then the x-axis coordinate is changed.

[0194] Repeat the process of acquiring the foreign object pixel coordinate data corresponding to the first space side, and acquire the foreign object pixel coordinates corresponding to the second to sixth space sides respectively. Mark the pixel corresponding to each foreign object pixel coordinate in the foreign object monitoring space volume to obtain multiple abnormal pixel points. Use a feature cube to fill the foreign object pixel points and perform volume accumulation on the feature cube to obtain the foreign object region volume value.

[0195] It should be noted here that:

[0196] The feature cube involved here has a side length of one pixel.

[0197] The volume of the foreign object monitoring space is obtained, and the ratio of the foreign object area volume to the space volume is calculated to obtain the foreign object volume occupancy degree corresponding to the sample re-inspection area.

[0198] Repeat the process of obtaining the foreign matter volume occupancy degree corresponding to the sample re-inspection area, and obtain the foreign matter volume occupancy degree corresponding to the re-inspection area of ​​each channel to obtain the channel foreign matter re-inspection data.

[0199] The foreign object warning module provides foreign object warnings for the channel area based on the foreign object re-inspection data.

[0200] Specifically as follows:

[0201] Obtain channel foreign object re-inspection data, obtain the foreign object volume occupancy degree corresponding to each channel re-inspection area based on the channel foreign object re-inspection data, and set a foreign object volume occupancy benchmark interval.

[0202] If the foreign object volume encroachment is within the foreign object volume encroachment benchmark range, then a foreign object warning will be issued for the channel area; if the foreign object volume encroachment is not within the foreign object volume encroachment benchmark range, then no foreign object warning will be issued for the channel area.

[0203] It should be noted here that:

[0204] In this application, the upper limit of the foreign object volume encroachment benchmark interval is 100%, that is, the channel area is completely encroached by foreign objects. The historical channel areas that have been warned of foreign object encroachment are acquired, and the foreign object volume encroachment degree corresponding to each historical channel area is obtained. The foreign object volume encroachment degree with the smallest value is set as the lower limit of the foreign object volume encroachment benchmark interval.

[0205] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for monitoring a channel for foreign objects based on intelligent interpretation, characterized in that, The application relates to a channel foreign matter monitoring method. Step S1: remote sensing image collection is carried out on a channel monitoring area, a two-dimensional area intrusion degree of the channel monitoring area is analyzed according to a region remote sensing image, the channel monitoring area is screened, and channel foreign matter preliminary screening data are obtained; Step S2: region image collection is carried out on a channel re-inspection area according to the channel foreign matter preliminary screening data, a foreign matter monitoring space body is created, the channel foreign matter volume of the channel re-inspection area is monitored by using the foreign matter monitoring space body, foreign matter volume intrusion degrees are obtained, and channel foreign matter re-inspection data are obtained; The step S2 further comprises the following steps: Step S21: channel foreign matter preliminary screening data are obtained, and a sample re-inspection area is selected according to the channel foreign matter preliminary screening data; Step S22: channel foreign matter re-inspection is carried out on the sample re-inspection area, and foreign matter volume intrusion degrees corresponding to the sample re-inspection area are obtained according to a re-inspection result; Step S23: foreign matter volume intrusion degrees corresponding to each channel re-inspection area are obtained, and channel foreign matter re-inspection data are obtained; The step S22 further comprises the following steps: Step S221: a foreign matter monitoring space body is arranged in a space environment corresponding to the sample re-inspection area, and a three-dimensional coordinate system is created for the foreign matter monitoring space body, so that a foreign matter monitoring three-dimensional coordinate system is obtained; Step S222: space sides contained in the foreign matter monitoring space body are respectively set as a first space side to a sixth space side; Step S223: image collection is carried out on the first space side to the sixth space side, coordinate collection is carried out on effective foreign matter pixel points according to the collected images, and foreign matter pixel coordinates are obtained; Step S224: pixel points corresponding to the foreign matter pixel coordinates are marked in the foreign matter monitoring space body, the marked pixel points are filled by using a feature cube, volume accumulation is carried out on the filled feature cube, and a foreign matter area volume value is obtained; Step S225: volume values of the foreign matter monitoring space body are obtained, a space body volume value is obtained, a ratio of the foreign matter area volume value to the space body volume value is calculated, and a foreign matter volume intrusion degree is obtained; Step S3: channel area foreign matter early warning is carried out according to the channel foreign matter re-inspection data.

2. The smart interpretation based tunnel foreign object monitoring method according to claim 1, characterized in that, The step S1 further comprises the following steps: Step S11: a plurality of channel monitoring areas are obtained, channel remote sensing image collection is carried out on an abnormal monitoring area, and a plurality of region remote sensing images are obtained; Step S12: image analysis is carried out on the region remote sensing images, and a channel area abnormal intrusion degree corresponding to each channel monitoring area is obtained according to an analysis result; Step S13: an abnormal intrusion benchmark interval is set, if the channel area abnormal intrusion degree is in the abnormal intrusion benchmark interval, the corresponding channel monitoring area is set as a channel re-inspection area, if not, the corresponding channel monitoring area is set as a channel foreign matter-free area, and channel foreign matter preliminary screening data are obtained.

3. The smart interpretation based tunnel foreign object monitoring method according to claim 2, characterized in that, The step S12 further comprises the following steps: Step S121: a sample monitoring area is randomly selected in the abnormal monitoring area, and a region remote sensing image corresponding to the sample monitoring area is set as a sample region remote sensing image; Step S122: marking the abnormal monitoring area in the sample area remote sensing image to obtain a marked monitoring area, setting a pixel traversal point in the marked monitoring area, obtaining the pixel point where the pixel traversal point is located at the current time to obtain a real-time monitoring pixel point; Step S123: performing pixel depth analysis on the real-time monitoring pixel point, and obtaining the pixel depth deviation corresponding to the real-time monitoring pixel point; Step S124: using the pixel traversal point to traverse the pixel points in the marked monitoring area, and obtaining the pixel depth deviation corresponding to each pixel point, if the pixel depth deviation is greater than a pixel depth deviation reference value, the pixel point is divided into a foreign object pixel point, and if it is less than or equal to, the pixel point is divided into a channel pixel point; Step S125: performing aggregation analysis on the foreign object pixel points in the marked monitoring area to obtain a channel area abnormal occupation degree, and obtaining the channel area abnormal occupation degree corresponding to each abnormal monitoring area.

4. The smart interpretation based tunnel foreign object monitoring method according to claim 3, characterized in that, In the step S123, the following steps are further included: The sample area remote sensing image is subjected to gray scale processing to obtain a sample area gray scale image, a coordinate system is created for the marked monitoring area to obtain a gray area plane rectangular coordinate system, the coordinates of the real-time monitoring pixel point in the gray area plane rectangular coordinate system are obtained to obtain a feature monitoring coordinate; The obtained historical remote sensing images are subjected to gray scale processing to obtain a plurality of historical remote sensing gray scale images, a coordinate system is created for the historical remote sensing gray scale images to obtain a historical area plane rectangular coordinate system, the pixel point corresponding to the feature monitoring coordinate in the historical area plane rectangular coordinate system is obtained to obtain a plurality of historical non-foreign object pixel points, and a reasonable pixel depth preset interval is set accordingly; The real-time monitoring pixel depth is obtained, the difference between the real-time monitoring pixel depth and the reasonable pixel depth preset interval is calculated to obtain the pixel depth deviation corresponding to the real-time monitoring pixel point.

5. The smart interpretation based tunnel foreign object monitoring method as claimed in claim 3, wherein, In the step S125, the following steps are further included: Any sample abnormal pixel point in the marked monitoring area is selected, the distance between the sample abnormal pixel point and each foreign object pixel point is obtained, the minimum distance is set as the adjacent abnormal pixel distance corresponding to the sample abnormal pixel point; The adjacent abnormal pixel distance corresponding to each foreign object pixel point is obtained, an abnormal pixel reference distance is set, the number of foreign object pixel points with a neighboring abnormal pixel distance less than or equal to the abnormal pixel reference distance is counted to obtain an effective foreign object pixel point number value; The pixel point number value in the marked monitoring area is counted to obtain a region pixel point number value, the ratio of the effective foreign object pixel point number value to the region pixel point number value is calculated to obtain the channel area abnormal occupation degree corresponding to the sample monitoring area.

6. The smart interpretation based tunnel foreign object monitoring method as claimed in claim 1, wherein, In the step S223, the following steps are further included: Step S2231: setting an image acquisition terminal in front of the first space side, and naming the cameras included in the image acquisition terminal as a first terminal camera and a second terminal camera, respectively; Step S2232: image acquisition of the first space side is performed using the first terminal camera and the second terminal camera to obtain a first side image and a second side image; Step S2233: a sample foreign object pixel point is selected from the effective foreign object pixel points in the first side image, and binocular vision analysis is performed on the sample foreign object pixel point to obtain a first terminal linear distance; Step S2234: the first terminal linear distance corresponding to each effective foreign object pixel point is obtained, the linear distance of the first terminal camera and the first space side is obtained, a terminal side distance difference is obtained, if the effective foreign object pixel point is greater than or equal to the terminal side distance difference, the corresponding effective pixel point is set as an in-plane foreign object pixel point, and if it is less than the terminal side distance difference, the effective pixel point is set as an out-of-plane foreign object pixel point; Step S2235: the coordinate position of the first terminal camera in the foreign object monitoring three-dimensional coordinate system is obtained to obtain a first terminal coordinate position, and the coordinate position of the in-plane foreign object pixel point in the foreign object monitoring three-dimensional coordinate system is obtained according to the first terminal coordinate position to obtain a foreign object pixel coordinate.

7. The smart interpretation based tunnel foreign object monitoring method according to claim 6, characterized in that, In the step S2233, the following steps are further included: The sample foreign object pixel point in the first side image is marked as a first foreign object pixel position, the sample foreign object pixel point in the second side image is obtained, the first side image and the second side image are edge-aligned, and the relative position of the sample foreign object pixel point in the second side image is projected to the first side image to obtain a second foreign object pixel position; In the first side image, the linear distance of the first foreign object pixel position and the second foreign object pixel position is obtained to obtain a foreign object pixel point disparity; The optical center distance between the first terminal camera and the second terminal camera is obtained to obtain a terminal device optical center distance, and the focal length of the first terminal camera is obtained to obtain a terminal device focal length; The terminal device optical center distance, the terminal device focal length, and the foreign object pixel point disparity are calculated to obtain the first terminal linear distance.

8. The smart interpretation based tunnel foreign object monitoring method as claimed in claim 1, wherein, In the step S3, the following steps are further included: The channel foreign object re-inspection data is obtained, the foreign object volume intrusion degree corresponding to each channel re-inspection region is obtained according to the channel foreign object re-inspection data, and a foreign object volume intrusion reference interval is set; If the foreign object volume intrusion degree is in the foreign object volume intrusion reference interval, the channel region is warned of foreign objects, and if it is not in the foreign object volume intrusion reference interval, the channel region does not need to be warned of foreign objects.

9. A system for intelligent interpretation based tunnel foreign object detection, adapted to the method for intelligent interpretation based tunnel foreign object detection according to any one of claims 1 to 8, characterized in that The monitoring system comprises: An image acquisition module: remote sensing image acquisition is performed on the channel monitoring region, region two-dimensional intrusion degree analysis is performed on the channel monitoring region according to the region remote sensing image, the channel monitoring region is screened to obtain channel foreign object preliminary screening data; A foreign object re-inspection module: region image acquisition is performed on the channel re-inspection region according to the channel foreign object preliminary screening data, and a foreign object monitoring space body is created, the channel foreign object volume of the channel re-inspection region is monitored using the foreign object monitoring space body, the foreign object volume intrusion degree corresponding to each channel re-inspection region is obtained to obtain channel foreign object re-inspection data; A foreign object warning module: the channel region is warned of foreign objects according to the channel foreign object re-inspection data.

Images