Method and device for treating dangerous points along a railway

Abstract

The application relates to a railway line danger point processing method and device, wherein the method comprises the following steps: acquiring railway line multi-temporal remote sensing image data and performing pretreatment and other operations thereon; preparing railway line ground object training sample data with marking information, and marking and preprocessing five main ground object elements; training a deep learning ground object detection model by using the ground object training sample data, extracting various railway line ground object elements, and converting the various railway line ground object elements into vector data matched with geographical coordinates; obtaining danger point elements by change detection, calculating the distance from the change elements to the railway, dividing the danger levels, and judging whether on-site exploration and hidden danger elimination are needed. Thus, the problems that the existing railway line danger point detection method has low detection precision and efficiency, and cannot meet the detection requirements of large-scale railway line danger points are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite remote sensing monitoring, and particularly relates to a railway line dangerous point processing method and device. BACKGROUND

[0002] Compared with unmanned aerial vehicle aerial photography, Google map and other technical means, high-resolution remote sensing images have wide coverage, high resolution, strong real-time performance, low cost and strong periodicity, and provide an effective technical means for rapid, dynamic and objective detection of railway line dangerous points.

[0003] At present, the detection of railway line dangerous points less uses remote sensing technology to detect the railway line dangerous points, but mainly relies on manual investigation by railway staff, and has the defects of long detection time, low efficiency, large consumption of manpower and material resources, subjective factor interference and the like, and it is difficult to comprehensively investigate the entire railway network, with the rapid development and expansion of the railway network, the railway project route becomes longer and passes through many complex terrain areas, and many places are inconvenient to reach, the traditional manual investigation cannot meet the detection requirements of large-scale railway line dangerous points, and problems are missed.

[0004] In summary, the existing railway line dangerous point detection method has low detection precision and efficiency, cannot meet the detection requirements of large-scale railway line dangerous points, and needs to be solved urgently. SUMMARY

[0005] The present application provides a railway line dangerous point processing method and device to solve the problems of low detection precision and efficiency of the existing railway line dangerous point detection method, and the inability to meet the detection requirements of large-scale railway line dangerous points.

[0006] The first aspect embodiment of the present application provides a railway line dangerous point processing method, comprising the following steps: acquiring multi-temporal remote sensing image data of a target railway line, and pre-processing and data registration operation on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data; collecting remote sensing image data of the target railway line, and labeling the remote sensing image data to construct a ground feature training sample data set of the target railway line, and training a pre-constructed ground feature detection model through the ground feature training sample data set; cropping the standard image data to obtain a plurality of target size standard image data, and inputting the plurality of target size standard image data into the trained ground feature detection model to generate ground feature prediction results of a plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data; coordinate matching and data conversion operation is performed on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, and change detection processing is performed on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data, and calculate the distance between each change element in the at least one change element and the target railway line, and determine the dangerous level of each change element according to the distance, to execute dangerous processing operation corresponding to each change element based on the dangerous level.

[0007] Optionally, in an embodiment of the present application, the multi-temporal remote sensing image data of the target railway line is acquired, and the multi-temporal remote sensing image data is pre-processed and data registration operation is performed to obtain the standard image data corresponding to the multi-temporal remote sensing image data, comprising: based on the preset detection interval, the multi-temporal remote sensing image data is cropped to obtain the multi-temporal remote sensing image data in the detection interval; data processing and radiation correction operation is performed on the multi-temporal remote sensing image data in the detection interval to obtain multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data; data registration operation is performed on the multi-temporal remote sensing processing data to generate the standard image data.

[0008] Optionally, in an embodiment of the present application, the remote sensing image data of the target railway line is collected, and the remote sensing image data is labeled to construct the ground object training sample data set of the target railway line, and the pre-constructed ground object detection model is trained through the ground object training sample data set, comprising: through a pre-set deep learning label making tool, a plurality of target ground objects in the remote sensing image data are labeled to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads and water bodies along the target railway line; the ground object training sample data set along the target railway line is constructed according to the label data and the remote sensing image data; the label data and the remote sensing image data in the ground object training sample data set are respectively cropped into a plurality of target size sub-region data, and a naming format uniform operation is performed on the plurality of target size sub-region data of the label data and the remote sensing image data; the pre-constructed ground object detection model is trained through the plurality of target size sub-region data after the naming format uniform operation.

[0009] Optionally, in an embodiment of the present application, the ground object prediction result is subjected to coordinate matching and data conversion operations to obtain ground object vector data corresponding to the ground object prediction result, and the ground object vector data is subjected to change detection processing to determine at least one change element corresponding to the ground object vector data, and the distance between each change element in the at least one change element and the target railway line is calculated, and the dangerous level of each change element is determined according to the distance, and based on the dangerous level, a dangerous processing operation corresponding to each change element is performed, comprising: obtaining actual geographic coordinates and pixel coordinates corresponding to the ground object prediction result, and matching the actual geographic coordinates and the pixel coordinates to generate coordinate matching data, and converting the coordinate matching data into a vector data format to obtain the ground object vector data; the ground object vector data corresponding to each target size standard image data is subjected to vector splicing to obtain spliced image vector data; the spliced image vector data is subjected to difference operation to obtain a difference result corresponding to the spliced image vector data, so as to determine the at least one change element according to the difference result; the distance between each change element and the target railway line is calculated, and a plurality of dangerous buffer zones are set according to the distance, so as to determine the dangerous level of each change element through the plurality of dangerous buffer zones; based on the pre-set railway safety operation requirement and the dangerous level, a target dangerous point element existing in the railway safety operation risk in the at least one change element is determined, and corresponding dangerous investigation operation is performed on the target dangerous point element.

[0010] The second aspect embodiment of the present application provides a railway line dangerous point processing device, comprising: a preprocessing module configured to obtain multi-temporal remote sensing image data of a target railway line, and perform preprocessing and data registration operations on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data; a training module configured to collect remote sensing image data of the target railway line, and label the remote sensing image data to construct a ground feature training sample data set of the target railway line, and train a pre-constructed ground feature detection model through the ground feature training sample data set; a prediction module configured to crop the standard image data to obtain a plurality of target size standard image data, and input the plurality of target size standard image data into the trained ground feature detection model to generate ground feature prediction results of a plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data; and an analysis module configured to perform coordinate matching and data conversion operations on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, perform change detection processing on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a dangerous level of each change element according to the distance, and perform dangerous processing operations corresponding to each change element based on the dangerous level.

[0011] Optionally, in an embodiment of the present application, the preprocessing module comprises: a first cropping unit configured to crop the multi-temporal remote sensing image data based on a preset detection interval to obtain multi-temporal remote sensing image data within the detection interval; a data processing unit configured to perform data processing and radiation correction operations on the multi-temporal remote sensing image data within the detection interval to obtain multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data; and a data registration unit configured to perform data registration operations on the multi-temporal remote sensing processing data to generate the standard image data.

[0012] Optionally, in an embodiment of the present application, the training module comprises: a labeling unit configured to label a plurality of target ground objects in the remote sensing image data by using a preset deep learning label making tool to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads and water bodies along the target railway line; a construction unit configured to construct ground object training sample data set along the target railway line according to the label data and the remote sensing image data; a second cropping unit configured to crop the label data and the remote sensing image data in the ground object training sample data set into a plurality of target size sub-region data respectively, and perform naming format unification operation on the plurality of target size sub-region data of the label data and the remote sensing image data; and a model processing unit configured to train the pre-constructed ground object detection model by using the plurality of target size sub-region data after the naming format unification operation.

[0013] Optionally, in an embodiment of the present application, the analysis module comprises: a matching unit configured to obtain actual geographic coordinates and pixel coordinates corresponding to the ground object prediction result, match the actual geographic coordinates and the pixel coordinates to generate coordinate matching data, and convert the coordinate matching data into a vector data format to obtain the ground object vector data; a splicing unit configured to perform vector splicing on the ground object vector data corresponding to each target size standard image data to obtain spliced image vector data; a difference unit configured to perform difference operation on the spliced image vector data to obtain a difference result corresponding to the spliced image vector data, and determine the at least one change element according to the difference result; a calculation unit configured to calculate distances between each change element and the target railway line, and set a plurality of dangerous buffer zones according to the distances to determine dangerous levels of each change element through the plurality of dangerous buffer zones; and a determination unit configured to determine a target dangerous point element with a railway safety operation risk in the at least one change element based on a preset railway safety operation requirement and the dangerous levels, and perform corresponding dangerous investigation operation on the target dangerous point element.

[0014] An electronic device is provided in a third aspect of embodiments of the present application, and includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor executes the program to implement the railway line danger point processing method as described in the above embodiments.

[0015] A computer readable storage medium is provided in a fourth aspect of embodiments of the present application, and the computer readable storage medium stores a computer program, and the program is executed by a processor to implement the railway line danger point processing method as described above.

[0016] The fifth aspect of the embodiment of the present application provides a computer program product, including a computer program executed to implement the railway line danger point processing method described above.

[0017] Therefore, the embodiments of the present application have the following beneficial effects:

[0018] The embodiments of the present application can obtain multi-temporal remote sensing image data of a target railway line, and perform preprocessing and data registration operations on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data; collect remote sensing image data of the target railway line, and label the remote sensing image data to construct a ground feature training sample data set of the target railway line, and train a pre-constructed ground feature detection model through the ground feature training sample data set; crop the standard image data to obtain a plurality of target size standard image data, and input the plurality of target size standard image data into the trained ground feature detection model to generate ground feature prediction results of a plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data; perform coordinate matching and data conversion operations on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, and perform change detection processing on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data, calculate the distance between each change element in the at least one change element and the target railway line, and determine the danger level of each change element according to the distance, and perform danger processing operations corresponding to each change element based on the danger level. The present application can significantly improve the efficiency and accuracy of railway line danger point detection, thereby providing important technical support for ensuring railway operation safety and playing a positive role in protecting people's life and property safety. Thus, the problems of low detection accuracy and efficiency of the existing railway line danger point detection method, which cannot meet the detection needs of large-scale railway line danger points, etc. are solved.

[0019] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description of embodiments, taken in conjunction with the accompanying drawings, in which:

[0021] Figure 1 A flowchart of a railway line danger point processing method according to an embodiment of the present application is provided.

[0022] Figure 2 A schematic diagram of setting a buffer zone using an ArcGIS Pro buffer zone tool according to an embodiment of the present application is provided.

[0023] Figure 3 A cropped 2021 detection area image map provided for an embodiment of the present application;

[0024] Figure 4 An execution logic diagram of a railway line danger point processing method provided for an embodiment of the present application;

[0025] Figure 5 An EasyFeature sample making tool diagram provided for an embodiment of the present application;

[0026] Figure 6 A 2021 building mask diagram processed by python provided for an embodiment of the present application;

[0027] Figure 7 A python code diagram of selected deep learning neural network related parameters provided for an embodiment of the present application;

[0028] Figure 8 A python code diagram of a cropping operation provided for an embodiment of the present application;

[0029] Figure 9 A python code diagram of matching predicted results with actual geographic coordinates provided for an embodiment of the present application;

[0030] Figure 10 A 2021 vector part effect diagram matched with image coordinates provided for an embodiment of the present application;

[0031] Figure 11 A python code diagram of a vector splicing operation to restore the original size vector after splicing the predicted result vector provided for an embodiment of the present application;

[0032] Figure 12 An ArcGIS Pro cropping tool vector difference diagram provided for an embodiment of the present application;

[0033] Figure 13 A building change detection diagram of a certain small area provided for an embodiment of the present application;

[0034] Figure 14 A diagram of calculating the distance from a feature to a rail using the ArcGIS Pro generated near neighbor table tool provided for an embodiment of the present application;

[0035] Figure 15 A 2021-2022 building change vector attribute diagram provided for an embodiment of the present application;

[0036] Figure 16 A schematic diagram of a new building element detection result of a certain piece of land provided by an embodiment of the present application;

[0037] Figure 17 A schematic diagram of a dangerous point verification result in a new building element detection result of a certain piece of land provided by an embodiment of the present application;

[0038] Figure 18 An example diagram of a railway line dangerous point processing device according to an embodiment of the present application;

[0039] Figure 19 A schematic diagram of the structure of an electronic device provided by an embodiment of the present application.

[0040] Among them, 10-railway line dangerous point processing device, 100-preprocessing module, 200-training module, 300-prediction module, 400-analysis module, 1901-memory, 1902-processor, 1903-communication interface. DETAILED DESCRIPTION

[0041] The embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by referring to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0042] A railway line danger point processing method and device are described below with reference to the accompanying drawings. To address the problems mentioned in the background art, the present application provides a railway line danger point processing method. In the method, multi-temporal remote sensing image data of a target railway line is obtained, and the multi-temporal remote sensing image data is preprocessed and data registration is performed to obtain standard image data corresponding to the multi-temporal remote sensing image data. Remote sensing image data of the target railway line is collected, and the remote sensing image data is labeled to construct a ground feature training sample data set of the target railway line, and a ground feature detection model is trained by the ground feature training sample data set. The standard image data is cropped to obtain a plurality of target size standard image data, and the plurality of target size standard image data is input into the trained ground feature detection model to generate ground feature prediction results of a plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data. Coordinate matching and data conversion operations are performed on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, and change detection processing is performed on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data. The distance between each change element in the at least one change element and the target railway line is calculated, and the danger level of each change element is determined according to the distance. Based on the danger level, a danger processing operation corresponding to each change element is performed. The present application can significantly improve the efficiency and accuracy of railway line danger point detection, thereby providing important technical support for ensuring railway operation safety and playing an active role in protecting people's life and property safety. Thus, the problems of low detection accuracy and efficiency of existing railway line danger point detection methods, which cannot meet the detection needs of large-scale railway line danger points, are solved.

[0043] Specifically, Figure 1 A flowchart of a railway line danger point processing method provided by an embodiment of the present application is shown in FIG. 1.

[0044] As Figure 1 shown, the railway line danger point processing method includes the following steps:

[0045] In step S101, multi-temporal remote sensing image data of a target railway line is obtained, and preprocessing and data registration operations are performed on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data.

[0046] Embodiments of the present application can first obtain multi-temporal remote sensing image data of a target railway line, and according to actual needs, the multi-temporal remote sensing image data containing the railway line region is cropped along the detection interval to obtain remote sensing images within the detection range, thereby reducing the amount of data processed and improving processing efficiency. Second, the obtained multi-temporal remote sensing image data is preprocessed and registered to ensure its spatial consistency for subsequent differential processing.

[0047] Optionally, in an embodiment of the present application, multi-temporal remote sensing image data along the target railway is acquired, and the multi-temporal remote sensing image data is preprocessed and data registration is performed to obtain standard image data corresponding to the multi-temporal remote sensing image data, including: based on a preset detection interval, the multi-temporal remote sensing image data is cropped to obtain multi-temporal remote sensing image data in the detection interval; the multi-temporal remote sensing image data in the detection interval is subjected to data processing and radiation correction to obtain multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data; and the multi-temporal remote sensing processing data is subjected to data registration to generate the standard image data.

[0048] In actual execution, the embodiment of the present application can acquire three images of a certain section of the Wuhan-Xianning railway in Wuhan City, China from 2021 to 2023, as shown in FIG. 1, taking the image in 2021 as an example, the embodiment of the present application can first use the buffer tool in ArcGIS Pro software to set a buffer zone of one kilometer away from the rail vector, and then perform registration, denoising, and cropping according to the buffer zone to obtain images along the left and right sides of the rail about one kilometer, as shown in FIG. 2, the specific steps are as follows: Figure 2 Figure 3

[0049] Step 1, according to the acquired rail line feature Shp file, the embodiment of the present application uses the buffer tool in ArcGIS Pro to set the input feature as the rail line feature and the buffer distance as one kilometer in linear units to output a strip-shaped buffer zone layer;

[0050] Step 2, according to the specific conditions of the image, use the geographic registration tool in ArcGIS Pro to perform geographic registration on the multi-temporal image, thereby preventing the subsequent change detection from being disturbed by the overall deviation of the geographic position; then use the filter tool to perform denoising processing on the image, thereby improving the image quality and facilitating subsequent deep learning interpretation, if the image needs radiation correction, ENVI or Erdas Imagine can be used for correction;

[0051] Step 3, finally use the mask extraction tool in ArcGIS Pro to set the input raster as the multi-temporal remote sensing image and the feature mask data as the strip-shaped buffer zone layer obtained in step 1 to realize the cropping of the detection area image one kilometer away from the rail.

[0052] After that, the embodiment of the present application can preprocess the cropped multi-temporal remote sensing image data by image denoising, enhancement, radiation correction, etc., and then register the multi-temporal remote sensing image data of different times and different sources, thereby providing reliable data guidance and basis for subsequent detection of dangerous points along the railway.

[0053] ​​In step S102, remote sensing image data along the target railway is collected, and the remote sensing image data is labeled to construct a ground object training sample data set along the target railway, and a pre-constructed ground object detection model is trained through the ground object training sample data set.

[0054] Further, the embodiments of the present application also need to use professional deep learning label making tools such as EasyFeature to label five main ground object elements along the railway, including buildings, vegetation, bare land, roads, and water bodies, and focus on ground objects that may pose a danger to railway operation, to construct a ground object training sample data set along the railway; secondly, a series of pretreatments are performed on the training sample data set: the image and the label are both cropped into small regions of 512x512 pixels, and the image and the label file of each region are uniformly named; then, the embodiments of the present application can reasonably divide the training data set into a training set, a validation set, and a test set, which are used for training, validation, and evaluation of the model.

[0055] Further, as shown in Figure 4 the embodiments of the present application can select a suitable deep learning network model (i.e., a ground object detection model) and pre-train the ground object detection model, input the pre-processed and divided training sample data set into the ground object detection model for training, adjust the parameters of the ground object detection model according to the training of the validation set and the test set, and optimize the structure and performance of the ground object detection model.

[0056] Optionally, in an embodiment of the present application, remote sensing image data along a target railway is collected, and the remote sensing image data is labeled to construct a ground object training sample data set along the target railway, and a pre-constructed ground object detection model is trained through the ground object training sample data set, including: a pre-set deep learning label making tool is used to label a plurality of target ground objects in the remote sensing image data to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads, and water bodies along the target railway; a ground object training sample data set along the target railway is constructed according to the label data and the remote sensing image data; the label data and the remote sensing image data in the ground object training sample data set are respectively cropped into a plurality of target size sub-region data, and a naming format unification operation is performed on the plurality of target size sub-region data of the label data and the remote sensing image data; the pre-constructed ground object detection model is trained through the plurality of target size sub-region data after the naming format unification operation.

[0057] It should be noted that the embodiments of the present application can prepare a ground object training sample data along the railway with marked information, including labeling five main ground object elements, and pre-processing the sample data; as Figure 5As shown, the embodiments of the present application can use EasyFeature3 software to draw training samples of five substrates; as Figure 6 As shown, the embodiments of the present application can use the python related library function to extract the mask of the corresponding feature, and organize and arrange the data, divide the data set, and construct the feature training sample data set along the railway, and the specific implementation steps are as follows:

[0058] Step 1, use professional deep learning label making software to obtain training sample data, and the embodiments of the present application can use EasyFeature3 software to draw training sample vector, use a variety of vector editing tools such as drawing polygons, semi-automatic acquisition, boundary local redrawing, and polygon splitting to draw vectors, and obtain five types of features in 2021 image;

[0059] Step 2, a series of pretreatment is performed on the training sample data: using the opencv library and gdal library functions of python, the binary feature mask is obtained, and the remote sensing image data and the label are cut into small areas of 512x512 pixels, and the image and the label file of each area are uniformly named.

[0060] Step 3, use the data set division function provided by pytorch to divide the training data set, and the proportion of training set, test set and validation set can be set to 8:1:1;

[0061] Further, the embodiments of the present application can train a deep learning feature detection model, and extract various features along the railway, including model selection, model training, feature extraction, matching geographical coordinates and converting vectors. As Figure 7 As shown, the embodiments of the present application can select D-LinkNet34 semantic segmentation neural network as the feature extraction model, and use the model pre-trained by 0.1m high resolution data of Chang-Gan railway for pre-training, then input the training sample data set after preprocessing and division into the feature detection model for training, according to the training condition of the validation set and the test set, adjust the model parameters, and optimize the feature detection model, so as to greatly guarantee the detection reliability and generalization performance of the feature detection model.

[0062] In step S103, the standard image data is cropped to obtain a plurality of target size standard image data, and the plurality of target size standard image data is input into the trained feature detection model to generate a plurality of target features corresponding to each target size standard image data in the plurality of target size standard image data.

[0063] In step S104, coordinate matching and data conversion operations are performed on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, and change detection processing is performed on the ground object vector data to determine at least one change element corresponding to the ground object vector data, and the distance between each change element in the at least one change element and the target railway line is calculated, and the danger level of each change element is determined according to the distance, so as to perform danger processing operation corresponding to each change element based on the danger level.

[0064] Further, after the training accuracy of the ground object detection model reaches the expectation, the embodiments of the present application can first crop the multi-temporal remote sensing image data along the railway line to 512x512 pixels (i.e. the target size) to obtain a plurality of target size standard image data, and then input the plurality of target size standard image data into the trained ground object detection model for prediction to extract the corresponding five main ground objects; match the pixel coordinates of the prediction result with the actual geographic coordinates to ensure the accuracy of the geographic location of the prediction result, convert the matched geographic coordinate data into vector data format for subsequent analysis and processing.

[0065] It can be understood that, since the multi-temporal remote sensing image data along the railway line is cropped to 512x512 pixels during prediction, the embodiments of the present application can first perform vector splicing to combine the ground object vectors in scattered small areas into vectors of the entire area, and perform difference operation on the spliced vector data to compare the differences of the ground objects in two time periods to identify newly added, disappeared or changed ground object elements; then, the embodiments of the present application perform post-processing on the detected change patches according to the difference results, so as to remove noise and irrelevant information to obtain accurate change elements.

[0066] In addition, the embodiments of the present application can use geographic information system tools or distance calculation algorithms to calculate the distance of each change element to the nearest railway; according to the distance between the change element and the railway, a buffer zone with different radii is set to represent the range of different danger levels, and the change elements are divided into different danger levels according to the distance from the railway, such as high, medium and low danger levels; finally, the embodiments of the present application can judge whether the change element may pose a threat to the safe operation of the railway in combination with the requirements of the safe operation of the railway and the ground object category of the change element;

[0067] Thus, the embodiments of the present application can identify the dangerous points that may cause danger to the safe operation of the railway by using the acquired multi-temporal remote sensing image data along the railway, using deep learning technology to extract and change detect various ground objects along the railway, thereby providing a reference for the railway department to maintain the safety of the railway, greatly solving the problems of long time-consuming, low efficiency, large consumption of manpower and material resources, subjective factor interference and the like in traditional manual detection, providing important technical support for ensuring the safe operation of the railway, playing an active role in protecting the safety of people's lives and property, saving resource costs, and promoting social stability.

[0068] Optionally, in an embodiment of the present application, the ground object prediction result is subjected to coordinate matching and data conversion operations to obtain ground object vector data corresponding to the ground object prediction result, and the ground object vector data is subjected to change detection processing to determine at least one change element corresponding to the ground object vector data, and the distance between each change element in the at least one change element and the target railway line is calculated, and the danger level of each change element is determined according to the distance, so as to perform a dangerous processing operation corresponding to each change element based on the danger level, including: acquiring actual geographic coordinates and pixel coordinates corresponding to the ground object prediction result, and matching the actual geographic coordinates and the pixel coordinates to generate coordinate matching data, and converting the coordinate matching data into a vector data format to obtain the ground object vector data; performing vector splicing on the ground object vector data corresponding to each target size standard image data to obtain spliced image vector data; performing a difference operation on the spliced image vector data to obtain a difference result corresponding to the spliced image vector data, so as to determine at least one change element according to the difference result; calculating the distance between each change element and the target railway line, and setting a plurality of dangerous buffer zones according to the distance, so as to determine the danger level of each change element through the plurality of dangerous buffer zones; determining a target dangerous point element that exists in at least one change element that has a risk of safe operation of the railway based on a preset requirement for safe operation of the railway and the danger level, and performing a corresponding dangerous investigation operation on the target dangerous point element.

[0069] Specifically, as shown in Figure 8 , the embodiments of the present application can perform the above image data cropping operation on the remote sensing images of 2022 and 2023; as shown in Figure 9 , the multi-period data is intelligently interpreted by using the trained ground object detection model, the pixel coordinates of the prediction result are matched with the actual geographic coordinates, the matched geographic coordinate data is converted into a vector data format, and finally the multi-period ground object element vector matched with the image coordinates is obtained, as shown in Figure 10 .

[0070] After the training accuracy of the ground object detection model reaches the expectation, the embodiment of the present application can use the cv2.copyMakeBorder function in opencv to first crop the image data of 2022 and 2023 to 512x512 pixels, and then use the trained ground object detection model to predict and extract the corresponding five main ground objects. Further, the embodiment of the present application can use the GetRasterBand function and the GetGeoTransform function in the Gdal library to match the pixel coordinates of the prediction results with the actual geographic coordinates, ensure the accuracy of the geographic location of the prediction results, and convert the matched geographic coordinate data into vector data format.

[0071] Further, the embodiment of the present application can obtain dangerous point elements through vector splicing, vector difference and change polygon post-processing change detection; as shown in Figure 11 As shown in Figure 12 The embodiment of the present application uses a cropping tool to obtain change element polygons through vector difference of two periods of vectors that need to be compared, and then performs post-processing to remove pseudo changes caused by angles, light or boundaries, etc., to obtain real ground object change vectors (i.e., ground object vector data); as shown in Figure 13 As shown in

[0072] Step 1, use the paste, crop and other functions in the opencv library to loop and splice the predicted vectors into the same vector file;

[0073] Step 2, load each type of ground object multi-period vector into ArcGIS Pro, use the cropping tool to input the two periods of vectors that need to be compared as the input elements and the clipping elements, and output different change element polygons;

[0074] Step 3, use the field lookup tool in the attribute table to set an area threshold to remove smaller changes, i.e., to filter out pseudo changes caused by angles, light, boundaries, etc., to obtain real ground object change vectors.

[0075] In addition, the embodiment of the present application also needs to calculate the distance from the change elements to the railway, and divide the danger level according to the distance to determine whether it is necessary to explore and eliminate hidden dangers in the field. As shown in Figure 14 The embodiment of the present application can generate a near neighbor table tool to calculate the distance from each change element in the ground object change vector to the track, and then use a buffer tool to set buffer zones of different danger levels. Finally, the embodiment of the present application combines the requirements of railway safe operation and the ground object category of the change element to determine whether the change element may pose a threat to the safety of railway operation; as shown in Figure 15As shown in the NEAR DIST column in the change element attribute table, each change element is arranged in ascending order of distance from the track, and the danger level of the dangerous point can be intuitively understood, and the dangerous point can be located in the ArcGIS Pro view by clicking the element in the attribute table. Figure 15 It can be seen that the distance between the dangerous element and the track is 53.7 m, and the danger level belongs to medium, and the position of the change element can be located in the ArcGIS Pro view by clicking the element in the attribute table. Figure 16 As shown, the embodiment of the application detects that part of the building elements are newly built in the block of vacant land, and the image comparison is viewed, as shown. Figure 17 As shown, the dangerous element is a newly built color steel house.

[0076] Specifically, the specific process of determining the danger point and the danger level according to the distance in the embodiment of the application is as follows:

[0077] 1. Load the track vector and the change element vector into ArcGIS Pro software, and then use the near neighbor table tool to calculate the vertical distance from each vector in the change element vector to the track, and add this distance to the attribute field of the vector;

[0078] 2. Use the buffer tool to set 0-20 m, 20-50 m and 50-100 m as high, medium and low danger levels respectively, and calculate the risk level of each change element according to the coordinates;

[0079] 3. According to the requirements of safe operation of railway and the category of change element itself, compare two images to determine whether the change element may pose a threat to the safe operation of the railway, such as a dangerous point (i.e. a target dangerous point element), and perform corresponding dangerous investigation operation on the target dangerous point element.

[0080] In summary, the embodiment of the application can clearly and intuitively display the position of the dangerous point and the distance from the dangerous point to the track, and has clear boundary vector and attribute information. In addition, the embodiment of the application can also classify the danger level of the possible dangerous point, and the detection method is real and reliable, and has strong operability, and the result of change detection and feature interpretation is good, thereby verifying the effectiveness and reliability of the railway along the dangerous point processing method of the application.

[0081] According to the railway line danger point processing method provided in the embodiment of the present application, the multi-temporal remote sensing image data of the target railway line is acquired, and the multi-temporal remote sensing image data is preprocessed and data registration is performed to obtain the standard image data corresponding to the multi-temporal remote sensing image data; the remote sensing image data of the target railway line is collected, and the remote sensing image data is labeled to construct the ground object training sample data set of the target railway line, and the ground object detection model constructed in advance is trained through the ground object training sample data set; the standard image data is cropped to obtain a plurality of target size standard image data, and the plurality of target size standard image data is input into the trained ground object detection model to generate the ground object prediction result of a plurality of target ground objects corresponding to each target size standard image data in the plurality of target size standard image data; coordinate matching and data conversion are performed on the ground object prediction result to obtain the ground object vector data corresponding to the ground object prediction result, and the change detection processing is performed on the ground object vector data to determine at least one change element corresponding to the ground object vector data, and the distance between each change element in the at least one change element and the target railway line is calculated, and the danger level of each change element is determined according to the distance, so that the danger processing operation corresponding to each change element is performed based on the danger level. The present application can significantly improve the efficiency and accuracy of railway line danger point detection, thereby providing important technical support for ensuring railway operation safety and playing an active role in protecting people's life and property safety.

[0082] Secondly, the railway line danger point processing device according to the embodiment of the present application is described with reference to the accompanying drawings.

[0083] Figure 18 is a block schematic diagram of the railway line danger point processing device of the embodiment of the present application.

[0084] As shown in Figure 18 , the railway line danger point processing device 10 comprises a preprocessing module 100, a training module 200, a prediction module 300 and an analysis module 400.

[0085] The preprocessing module 100 is configured to acquire the multi-temporal remote sensing image data of the target railway line, and perform preprocessing and data registration operation on the multi-temporal remote sensing image data to obtain the standard image data corresponding to the multi-temporal remote sensing image data.

[0086] The training module 200 is configured to collect the remote sensing image data of the target railway line, and label the remote sensing image data to construct the ground object training sample data set of the target railway line, and train the ground object detection model constructed in advance through the ground object training sample data set.

[0087] The prediction module 300 is configured to crop the standard image data to obtain a plurality of target size standard image data, and input the plurality of target size standard image data into the trained ground object detection model to generate a plurality of ground object prediction results corresponding to a plurality of target ground objects in each target size standard image data.

[0088] The analysis module 400 is configured to perform coordinate matching and data conversion operations on the ground object prediction results to obtain ground object vector data corresponding to the ground object prediction results, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, and determine a danger level of each change element according to the distance, so as to perform a danger processing operation corresponding to each change element based on the danger level.

[0089] Optionally, in an embodiment of the present application, the preprocessing module 100 comprises a first cropping unit, a data processing unit and a data registration unit.

[0090] The first cropping unit is configured to crop the multi-temporal remote sensing image data based on a preset detection interval to obtain multi-temporal remote sensing image data in the detection interval.

[0091] The data processing unit is configured to perform data processing and radiation correction operations on the multi-temporal remote sensing image data in the detection interval to obtain multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data.

[0092] The data registration unit is configured to perform data registration operations on the multi-temporal remote sensing processing data to generate standard image data.

[0093] Optionally, in an embodiment of the present application, the training module 200 comprises a labeling unit, a construction unit, a second cropping unit and a model processing unit.

[0094] The labeling unit is configured to label a plurality of target ground objects in the remote sensing image data by using a preset deep learning label making tool to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads and water bodies along the target railway line.

[0095] The construction unit is configured to construct ground object training sample data set along the target railway line according to the label data and the remote sensing image data.

[0096] The second cropping unit is configured to respectively crop the label data and the remote sensing image data in the ground object training sample data set into a plurality of target size sub-region data, and perform naming format uniform operation on the plurality of target size sub-region data of the label data and the remote sensing image data.

[0097] The model processing unit is configured to train a pre-constructed ground object detection model by performing the sub-region data of the plurality of target sizes after the naming format unification operation.

[0098] Optionally, in an embodiment of the present application, the analysis module 400 comprises a matching unit, a splicing unit, a difference unit, a calculation unit and a determination unit.

[0099] The matching unit is configured to obtain actual geographic coordinates and pixel coordinates corresponding to the ground object prediction result, match the actual geographic coordinates and the pixel coordinates to generate coordinate matching data, and convert the coordinate matching data into a vector data format to obtain the ground object vector data.

[0100] The splicing unit is configured to perform vector splicing on the ground object vector data corresponding to each target size standard image data to obtain spliced image vector data.

[0101] The difference unit is configured to perform a difference operation on the spliced image vector data to obtain a difference result corresponding to the spliced image vector data, and determine at least one change element according to the difference result.

[0102] The calculation unit is configured to calculate a distance between each change element and the target railway line, set a plurality of danger buffer zones according to the distance, and determine a danger level of each change element through the plurality of danger buffer zones.

[0103] The determination unit is configured to determine a target danger point element that has a railway safety operation risk in the at least one change element based on a preset railway safety operation requirement and the danger level, and perform a corresponding danger investigation operation on the target danger point element.

[0104] It should be noted that the above description of the railway line danger point processing method embodiment is also applicable to the railway line danger point processing device of this embodiment, which will not be described here.

[0105] The railway line danger point processing device provided by the embodiment of the present application comprises a preprocessing module, which is used to acquire multi-temporal remote sensing image data of a target railway line, and performs preprocessing and data registration operations on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data; a training module, which is used to collect remote sensing image data of the target railway line, and label the remote sensing image data to construct a ground object training sample data set of the target railway line, and train a ground object detection model constructed in advance through the ground object training sample data set; a prediction module, which is used to crop the standard image data to obtain a plurality of target size standard image data, and input the plurality of target size standard image data into the trained ground object detection model to generate ground object prediction results of a plurality of target ground objects corresponding to each target size standard image data in the plurality of target size standard image data; and an analysis module, which is used to perform coordinate matching and data conversion operations on the ground object prediction results to obtain ground object vector data corresponding to the ground object prediction results, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform a danger processing operation corresponding to each change element based on the danger level. The present application can significantly improve the efficiency and accuracy of railway line danger point detection, thereby providing important technical support for ensuring railway operation safety and playing an active role in protecting people's life and property safety.

[0106] Figure 19 The structure schematic diagram of the electronic device provided by the embodiment of the present application is shown. The electronic device can comprise:

[0107] The memory 1901, the processor 1902 and the computer program stored in the memory 1901 and executable on the processor 1902.

[0108] The processor 1902 implements the railway line danger point processing method provided in the above embodiment when executing the program.

[0109] Further, the electronic device further comprises:

[0110] The communication interface 1903 is used for communication between the memory 1901 and the processor 1902.

[0111] The memory 1901 is used to store the computer program executable on the processor 1902.

[0112] The memory 1901 can contain a high-speed RAM memory, and can also include a non-volatile memory, such as at least one disk memory.

[0113] If the memory 1901, the processor 1902 and the communication interface 1903 are implemented independently, the communication interface 1903, the memory 1901 and the processor 1902 can be connected with each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 19 Only one thick line is used to represent the bus in the figure, but it does not mean that there is only one bus or only one type of bus.

[0114] Optionally, in a specific implementation, if the memory 1901, the processor 1902 and the communication interface 1903 are integrated on a chip, the memory 1901, the processor 1902 and the communication interface 1903 can complete communication between each other through an internal interface.

[0115] The processor 1902 can be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0116] The embodiments of the present application also provide a computer readable storage medium, which has a computer program stored thereon, and the program is executed by a processor to implement the railway line danger point processing method.

[0117] The embodiments of the present application also provide a computer program product, which includes a computer program, and the computer program is executed to implement the railway line danger point processing method.

[0118] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. In addition, the usage of "N" means at least two, for example, two, three or the like, unless explicitly stated otherwise.

[0119] Furthermore, the terms "first", "second", or the like, are used merely as a designation of certain elements or features, and do not imply or connote relative importance or a specific order of categorization of the indicated features. Accordingly, features described as "first" or "second" can be explicitly or implicitly included in at least one of the features. In the description of the application, the term "N" means at least two, for example, two, three, etc., unless explicitly stated otherwise.

[0120] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments which can be managed as one or more modules, segments, or portions of code which include one or more executable instructions for implementing specific logic functions or steps, and alternate implementations are possible. In some embodiments, the processes and methods described can be executably encoded on a machine- readable medium in a data signal embodied in an electromagnetic signal, a wireless signal, or a propagated signal.

[0121] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of instructions to implement logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a computer- readable storage medium or a computer-readable signal medium. The computer- readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical connections), a portable computer diskette (a magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for example, optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0122] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware and in another embodiment, any of the following technologies, known in the art, or their combinations can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0123] Those skilled in the art can understand that all or part of the steps carried out by the above-mentioned embodiments can be completed by programs instructing related hardware, and the programs can be stored in a computer-readable storage medium. When the programs are executed, one or a combination of the steps of the method embodiments is included.

[0124] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0125] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for handling dangerous points along a railway line, characterized in that, The method comprises the following steps: acquiring multi-temporal remote sensing image data along a target railway line, and performing preprocessing and data registration operations on the multi-temporal remote sensing image data to obtain standard image data corresponding to the multi-temporal remote sensing image data; collecting remote sensing image data along the target railway line, and labeling the remote sensing image data to construct a ground feature training sample data set along the target railway line, and training a pre-constructed ground feature detection model through the ground feature training sample data set; cropping the standard image data to obtain a plurality of target size standard image data, and inputting the plurality of target size standard image data into the trained ground feature detection model to generate ground feature prediction results of a plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data; performing coordinate matching and data conversion operations on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, and performing change detection processing on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data, and calculating the distance between each change element in the at least one change element and the target railway line, and determining the danger level of each change element according to the distance, to perform danger handling operations corresponding to each change element based on the danger level; wherein the coordinate matching and data conversion operations on the ground feature prediction results to obtain ground feature vector data corresponding to the ground feature prediction results, and performing change detection processing on the ground feature vector data to determine at least one change element corresponding to the ground feature vector data, and calculating the distance between each change element in the at least one change element and the target railway line, and determining the danger level of each change element according to the distance, to perform danger handling operations corresponding to each change element based on the danger level, comprises: acquiring actual geographic coordinates and pixel coordinates corresponding to the ground feature prediction results, and matching the actual geographic coordinates and the pixel coordinates to generate coordinate matching data, and converting the coordinate matching data into a vector data format to obtain the ground feature vector data; performing vector splicing on the ground feature vector data corresponding to each target size standard image data to obtain spliced image vector data; performing a difference operation on the spliced image vector data to obtain a difference result corresponding to the spliced image vector data, to determine the at least one change element according to the difference result; calculating the distance between each change element and the target railway line, and setting a plurality of danger buffer zones according to the distance, to determine the danger level of each change element through the plurality of danger buffer zones; based on the preset railway safety operation requirements and the danger level, determining a target dangerous point element in the at least one change element that has a railway safety operation risk, and performing corresponding danger investigation operations on the target dangerous point element.

2. The method of claim 1, wherein, The acquisition target railway along the multi-temporal remote sensing image data, and the multi-temporal remote sensing image data is preprocessed and data registration operation, to obtain the multi-temporal remote sensing image data corresponding to the standard image data, comprising: Based on the preset detection interval, the multi-temporal remote sensing image data is cropped to obtain the multi-temporal remote sensing image data in the detection interval; The multi-temporal remote sensing image data in the detection interval is processed and radiometrically corrected to obtain the multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data; The multi-temporal remote sensing processing data is registered to generate the standard image data.

3. The method of claim 2, wherein, The acquisition target railway along the remote sensing image data, and the remote sensing image data is labeled to construct the target railway along the ground feature training sample data set, and the ground feature detection model is trained by the ground feature training sample data set, comprising: Through the preset deep learning label making tool, a plurality of target ground features in the remote sensing image data are labeled to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground features include buildings, vegetation, bare land, roads and water bodies along the target railway; According to the label data and the remote sensing image data, the ground feature training sample data set along the target railway is constructed; The label data and the remote sensing image data in the ground feature training sample data set are respectively cropped into a plurality of target size sub-region data, and the label data and the remote sensing image data of the plurality of target size sub-region data are executed naming format uniform operation; The pre-constructed ground feature detection model is trained by the plurality of target size sub-region data after the naming format uniform operation.

4. A device for handling dangerous points along a railway line, characterized in that Comprising: The preprocessing module is used for acquiring the multi-temporal remote sensing image data along the target railway, and the multi-temporal remote sensing image data is preprocessed and data registration operation, to obtain the multi-temporal remote sensing image data corresponding to the standard image data; The training module is used for collecting the remote sensing image data along the target railway, and the remote sensing image data is labeled to construct the ground feature training sample data set along the target railway, and the ground feature detection model is trained by the ground feature training sample data set; The prediction module is used for cropping the standard image data to obtain a plurality of target size standard image data, and inputting the plurality of target size standard image data into the trained ground feature detection model to generate the ground feature prediction result of the plurality of target ground features corresponding to each target size standard image data in the plurality of target size standard image data. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The analysis module is configured to perform coordinate matching and data conversion on the ground object prediction result to obtain ground object vector data corresponding to the ground object prediction result, perform change detection processing on the ground object vector data to determine at least one change element corresponding to the ground object vector data, calculate a distance between each change element in the at least one change element and the target railway line, determine a danger level of each change element according to the distance, and perform danger processing corresponding to each change element based on the danger level. The preprocessing module includes:

5. The apparatus of claim 4, wherein, The first cropping unit is configured to crop the multi-temporal remote sensing image data based on a preset detection interval to obtain multi-temporal remote sensing image data in the detection interval. The data processing unit is configured to perform data processing and radiation correction on the multi-temporal remote sensing image data in the detection interval to obtain multi-temporal remote sensing processing data corresponding to the multi-temporal remote sensing image data. The data registration unit is configured to perform data registration on the multi-temporal remote sensing processing data to generate the standard image data. The training module includes:

6. The apparatus of claim 5, wherein, The labeling unit is configured to label a plurality of target ground objects in the remote sensing image data by using a preset deep learning label making tool to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads, and water bodies along the target railway line. The labeling unit is configured to label a plurality of target ground objects in the remote sensing image data by using a preset deep learning label making tool to obtain label data corresponding to the remote sensing image data, wherein the plurality of target ground objects include buildings, vegetation, bare land, roads, and water bodies along the target railway line. A constructing unit is configured to construct the ground object training sample dataset of the target railway line according to the label data and the remote sensing image data; A second cropping unit is configured to crop the label data and the remote sensing image data in the ground object training sample dataset into a plurality of sub-region data of target sizes respectively, and perform a naming format unification operation on the plurality of sub-region data of target sizes of the label data and the remote sensing image data. A model processing unit is configured to train the pre-constructed ground object detection model by using the plurality of sub-region data of target sizes after the naming format unification operation.

7. An electronic device, comprising: Comprise: A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the railway line danger point processing method according to any one of claims 1-3.

8. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the railway line danger point processing method according to any one of claims 1-3.

9. A computer program product comprising a computer program, characterized in that, The computer program is executed to implement the railway line danger point processing method according to any one of claims 1-3.

Images