A railway external environment safety hidden danger identification method based on remote sensing interpretation library

Abstract

The application discloses a railway external environment safety hidden danger identification method based on a remote sensing interpretation library, and comprises the following steps: S1, formulating coding rules of interpretation library images according to a railway external environment safety hidden danger classification system; S2, taking a single safety hidden danger ground object as an object, making field shooting photos, satellite images, aerial photographs and structure diagrams of the object, and uniformly naming each reference interpretation diagram according to the coding rules of step S1; S3, establishing an interpretation reference image database: selecting a relational database to manage interpretation image related information and project information, including an image information management table and a project information management table, and establishing a connection relationship between the two tables through project coding; S4, managing interpretation reference image data; S5, identifying railway external environment safety hidden dangers by remote sensing; and S6, updating the remote sensing interpretation library. The method makes the remote sensing checking work of railway external environment safety hidden dangers more accurate, efficient and convenient.

Description

Technical Field

[0001] This invention belongs to the field of railway external environment management, specifically involving a method for identifying railway external environment safety hazards based on a remote sensing interpretation library. Background Technology

[0002] Human activities around railways can pose potential safety hazards to their normal operation. While the technical process for documenting, monitoring, and assisting in the remediation of safety hazards in the external environment along railway lines using remote sensing is relatively mature, including sample annotation for deep learning and document creation using aerial photogrammetry, the knowledge and experience gained from applying remote sensing technology in the remediation of railway external environmental safety hazards has not been systematically and standardized. This leads to new problems such as data silos, resource waste, and repetitive work.

[0003] While current machine recognition technology is convenient, fast, and highly automated, its accuracy in identifying safety hazards in the railway external environment does not meet the requirements for engineering applications, still requiring subsequent manual interpretation and editing. However, in the process of manually interpreting remote sensing data for railway external environmental safety hazards, inaccurate attribute interpretation and inconsistent delineation standards frequently occur. Furthermore, remote sensing interpreters lack reference standards and learning materials for identifying railway external environmental safety hazards, and have not yet established an intuitive prior knowledge base to assist in remote sensing interpretation. Summary of the Invention

[0004] To address the technical problems existing in known technologies, this invention provides a method for identifying safety hazards in the external environment of railways based on a remote sensing interpretation library.

[0005] Therefore, the present invention adopts the following technical solution:

[0006] A method for identifying safety hazards in the external environment of railways based on a remote sensing interpretation library includes:

[0007] S1, Remote sensing interpretation library image coding: Develop coding rules for interpretation library images based on the railway external environment safety hazard classification system;

[0008] S2, Create interpretation reference images: Taking a single safety hazard feature as the object, create on-site photographs, satellite images, aerial photographs and structural diagrams of the object. Each interpretation reference image is uniformly named according to the coding rules of step S1.

[0009] S3, Establish an interpretation reference image database: Select a relational database to manage the relevant information and project information of the interpretation reference images produced in S2, including an image information management table and a project information management table, and establish a connection relationship between the image information management table and the project information management table through project coding;

[0010] S4, Interpretation Reference Image Data Management: Based on the interpretation reference image database established in step S3, with the interpretation database image data as the core, the interpretation reference images produced in S2 are classified and stored in JPG format, and the railway external environment safety hazard monitoring project and interpretation reference image attribute information are managed in the form of a database.

[0011] S5, Remote sensing identification of safety hazards in the external environment of railways: High spatial resolution remote sensing images with a resolution better than 1m are used as the data source for extraction. The extraction range is within 1km on both sides of the railway mainline. Aerial photography image processing and satellite image preprocessing are carried out. Computer automatic extraction and manual interpretation and editing are used to extract safety hazards in the external environment of railways. When interpreting the safety hazards, the interpretation of safety hazard types refers to the data resources in the remote sensing interpretation library of safety hazards in the external environment of railways. Finally, a vector image of safety hazards in the external environment of railways is obtained.

[0012] S6, Remote Sensing Interpretation Library Update: After the completion of the new phase of remote sensing extraction of railway external environmental safety hazards, and in conjunction with field verification data, representative railway external environmental safety hazards are interpreted as follows: reference images are created according to step S2, interpretation library images are encoded according to step S1, and interpretation reference image data is stored in the library according to step S4, thus completing the expansion and correction of the railway external environmental safety hazard remote sensing interpretation library.

[0013] The remote sensing interpretation library image encoding in step S1 includes major category of security hazard, minor category of security hazard, acquisition time, project number, risk source number, data source type and image sequence number.

[0014] In step S2, for on-site photographs, the longitude, latitude, and altitude information of the shooting time and location are recorded; for satellite imagery, the safety hazard patches are centered within the output area, the satellite imagery scaling is adjusted to ensure the safety hazard patches are fully displayed without pixelation, a scale bar icon is introduced, and the satellite imagery acquisition time and the longitude and latitude information of the safety hazard center are recorded; for aerial photography, the production process is the same as that for satellite imagery production; for structural diagrams, they are converted to JPG format.

[0015] The image information management table mentioned in step S3 includes 11 fields: code, project code, major safety hazard category, minor safety hazard category, longitude, latitude, altitude, data source category, storage location, geographical location, and remarks; the project information management table includes 2 fields: project code and project name. Among them:

[0016] The project coding field serves as a unique identifier to encode the remote sensing identification project for safety hazards in the external environment of the railway, and is associated with the image information management table;

[0017] The project name field is used to record the naming of projects for remote sensing identification of safety hazards in the external environment of railways.

[0018] The image information management table is used to record the attribute information of each interpretation reference image, wherein:

[0019] The encoding field is used to record the decoding library image encoding described in step S1;

[0020] The project code field is used as a foreign key to connect to the project information management table;

[0021] The "Safety Hazard Category" field is used to record the category of railway external environment safety hazards to which the image belongs;

[0022] The safety hazard category field is used to record the railway external environment safety hazard category to which the image belongs;

[0023] The longitude field is used to record the center longitude coordinates of the associated safety hazard;

[0024] The latitude field is used to record the center latitude coordinates of the associated safety hazard;

[0025] The altitude field is used to record the center elevation of the associated safety hazard.

[0026] The data source category field is used to record the data source type of the interpretation reference image;

[0027] The storage location field is used to record the file storage path of the interpretation reference image;

[0028] The geographic location field is used to record the spatial location description of the associated safety hazard;

[0029] The remarks field is used to record the attribute names of other categories within the subcategory of railway external environmental safety hazards.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] This invention is based on a remote sensing interpretation library of railway external environmental safety hazards, which is composed of on-site photographs, satellite images, aerial photography, and structural diagrams. It unifies the remote sensing identification standards for railway external environmental safety hazards and applies them to railway external environmental safety hazard remote sensing monitoring projects, making the remote sensing investigation of railway external environmental safety hazards more accurate, efficient, and fast. Attached Figure Description

[0032] Figure 1 This is a flowchart of the railway external environment safety hazard identification method of the present invention;

[0033] Figure 2 This is a schematic diagram of the image encoding rules in this invention;

[0034] Figure 3 This is a schematic diagram illustrating the connection relationship between the image information management table and the project information management table in this invention;

[0035] Figure 4 A schematic diagram illustrating the category numbering for folder naming in this invention;

[0036] Figure 5 This is an example of creating the interpretation reference image coded as M22112201005s1 in this invention. Detailed Implementation

[0037] The identification method of the present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0038] The flowchart of the railway external environment safety hazard identification method based on remote sensing interpretation library of the present invention is as follows: Figure 1 As shown. The method includes: S1, encoding remote sensing interpretation library images; S2, creating interpretation reference images; S3, designing the interpretation reference image database; S4, managing interpretation reference image data; S5, remote sensing identification of safety hazards in the railway's external environment; and S6, updating the remote sensing interpretation library. Details are as follows:

[0039] S1, Image encoding of the decoding library:

[0040] The railway external environmental safety hazard classification system includes 18 major categories and 45 subcategories, as detailed in Table 1:

[0041] Table 1

[0042]

[0043]

[0044] The image encoding in the decoding library contains 14 characters, following the rules as follows: Figure 2 As shown. Wherein:

[0045] The first digit indicates the major category of safety hazard, and they are coded sequentially according to the English alphabet from A to Z. Currently, there are 18 major categories of safety hazards in the external environment of railways.

[0046] The second digit indicates the sub-category of safety hazards, numbered sequentially from 1 to 9. The "Other" sub-category included in this category is always numbered 9, but the specific type of safety hazard in the external environment of the railway must be clearly stated in the remarks.

[0047] The third to eighth digits represent the original data acquisition time, expressed as numbers. The third and fourth digits represent the last two digits of the year, the fifth and sixth digits represent the month, and the seventh and eighth digits represent the day. If the acquisition time is unclear, it is represented by 99999.

[0048] The ninth digit represents the project number, which is sequentially encoded using the numbers 1 through 9, a through z, A through Z.

[0049] The tenth and twelfth digits represent the risk source number, which is sequentially coded from 001 to 999.

[0050] The thirteenth character indicates the data source type, which is represented by the lowercase first letter of the English name of the image source type. Among them, satellite imagery is abbreviated as s; aerial photography is abbreviated as a; field photos are abbreviated as p; and drawing is abbreviated as d.

[0051] The fourteenth digit represents the image number of the higher-level ground features, which is sequentially coded using numbers 1 to 9.

[0052] The image encoding in the interpretation library is a unique string of names that corresponds to each image in the remote sensing interpretation library for railway external environmental safety hazards.

[0053] S2, Create interpretation reference images to standardize and unify the creation standards for reference images of various security risks in the interpretation library:

[0054] For a single safety hazard feature, produce on-site photographs, satellite imagery, aerial photography, and structural diagrams. Each interpretation reference image should be uniformly named according to the requirements of step S1. For on-site photographs, record the time, location, longitude, latitude, and altitude information of the shooting location. For satellite imagery, center the safety hazard feature within the output area, adjust the satellite image zoom to ensure the safety hazard feature is fully displayed without pixelation, introduce a scale bar icon, and record the satellite image acquisition time and the longitude and latitude information of the safety hazard center. For aerial photography, the production process is the same as for satellite imagery. For structural diagrams (schematic diagrams, design drawings), JPG image format conversion is required.

[0055] S3, Establish a decoding reference image database to store decoding library image data information and project information:

[0056] A relational database is used to manage the relevant information and project information of the interpretation reference images created by S2, including an image information management table and a project information management table. These two tables are linked through project codes, such as... Figure 3 As shown. The image information management table contains 11 fields: code, project code, major safety hazard category, minor safety hazard category, longitude, latitude, altitude, data source category, storage location, geographical location, and remarks; the project information management table contains 2 fields: project code and project name.

[0057] S4, Interpretation Reference Image Data Management:

[0058] Based on the interpretation reference image database established in step S3, the interpretation reference files and attribute information created in step S2 are managed using the interpretation library image data as the core, and the "database + file" format is used to realize the storage, retrieval and viewing functions of the remote sensing interpretation library.

[0059] Storage space is established according to the categories of railway external environmental safety hazards to store various remote sensing interpretation reference images in JPG format. The project information management table in the database records information on railway external environmental safety hazard remote sensing identification projects. The "Project Code" field uniquely identifies each project and is linked to the image information management table. The "Project Name" field records the naming of each project. The image information management table records the attribute information of each interpretation reference image. The "Code" field records the interpretation image code mentioned in S1, and the "Project Code" field is linked to the project information management table as a foreign key. The "Safety Hazard Category" field records the railway external environmental safety hazard category to which the image belongs. The "Major Category" field records the railway external environment safety hazard subcategory to which the image belongs; the "Longitude" field records the center longitude coordinates of the safety hazard; the "Latitude" field records the center latitude coordinates of the safety hazard; the "Altitude" field records the center altitude of the safety hazard; the "Data Source Category" field records the data source type of the reference interpretation image; the "Storage Location" field records the file storage path of the reference interpretation image; the "Geographic Location" field records the spatial location description of the safety hazard; and the "Remarks" field records the attribute names of other categories within the railway external environment safety hazard subcategory.

[0060] S5, Remote sensing identification of safety hazards in the external environment of railways:

[0061] In this step, information is extracted from remote sensing images of the railway perimeter that require external environmental remediation, referring to the remote sensing interpretation library. Specifically:

[0062] High spatial resolution remote sensing imagery with a resolution better than 1m was used as the data source for extraction. The extraction range was within 1km on both sides of the railway mainline. Aerial photogrammetry (image linkage, aerial triangulation, mosaicking and cropping) and satellite image preprocessing (radiometric correction, image fusion, orthorectification, mosaicking and cropping) were carried out. Computer-automated extraction and manual interpretation and editing were used to remotely extract safety hazards in the railway's external environment. During manual interpretation, the interpretation of safety hazards was based on data resources in the remote sensing interpretation library. Finally, vector plots of safety hazards in the railway's external environment were obtained.

[0063] S6, Remote Sensing Interpretation Library Update:

[0064] After the new phase of remote sensing extraction of railway external environmental safety hazards is completed, and combined with field verification data, representative railway external environmental safety hazards are interpreted and reference images are created according to step S2, interpreted library images are encoded according to step S1, and interpreted reference image data are stored in the library according to step S4, thus completing the expansion and correction of the railway external environmental safety hazard remote sensing interpretation library.

[0065] Example 1

[0066] A method for identifying safety hazards in the external environment of railways based on a remote sensing interpretation library includes:

[0067] S1, Image encoding of the decoding library:

[0068] according to Figure 2 The rules are encoded, for example: L 1 211230 1 013p 4 The following indicates: Hard floating objects—Corrugated steel tile houses—December 30, 2021—Project No. 1 [Wuhan-Guangzhou Railway (Changsha Section, Hunan)]—Safety Hazard No. 013—On-site photos—Photo 4; L 1 211220 1 013s 1 The image indicates: Hard floating objects—Corrugated steel roof houses—December 20, 2021—Project No. 1 [Wuhan-Guangzhou Railway (Changsha, Hunan Section)]—Safety Hazard No. 013—Satellite image—Image 1; L 1 2004201 013a 1 The image indicates: Hard floating objects—Corrugated steel roof houses—April 20, 2020—Project No. 1 [Wuhan-Guangzhou Railway (Changsha, Hunan Section)]—Safety Hazard No. 013—Aerial photography—Image 1; L 1 220104 1 013d 1 , indicating: Hard floating objects - Corrugated steel roof - January 4, 2022 - Project No. 1 [Wuhan-Guangzhou Railway (Changsha Section, Hunan)] - Safety Hazard No. 013 - Structural Diagram - Page 1.

[0069] S2, Create an interpretation reference image:

[0070] Taking the creation of image M22112201005s1 (light floating objects - plastic greenhouse - satellite imagery) as an example, GIS software was used for mapping. A4 paper size (210 mm long, 297 mm wide) and a resolution of 300 dpi were selected as the output standard. The safety hazard patches were centered, and the scale bar icon was placed in the lower center of the output area. The satellite image acquisition time, longitude, latitude, and altitude of the center of the safety hazard features were recorded. Figure 5 As shown.

[0071] S3, Establish a database of interpretation reference images:

[0072] The image information management table and project information management table are constructed based on the PostgreSQL database. The table structure design is shown in Table 2 and Table 3:

[0073] Table 2 Image Information Management Table

[0074] Field Name Data types Field size / bytes Is it necessary? coding varchar 14 yes Project Code char 1 no Major categories of safety hazards varchar 50 yes Safety Hazard Subcategories varchar 50 yes longitude float8 53 yes latitude float8 53 yes altitude float8 53 no Data source categories varchar 255 yes Storage location varchar 255 yes Geographical location varchar 255 no Remark varchar 12 no

[0075] Table 3 Project Information Management Table

[0076] Field Name Data types Field size / bytes Is it necessary? Project Code char 1 yes Project Name varchar 255 yes

[0077] S4, Interpretation Reference Image Data Management:

[0078] Based on the classification (major and minor categories) of railway external environmental safety hazards, folders will be established to store reference images for interpreting safety hazards. The folder names will adopt the category numbers designed in S1, in the following format: Figure 4 As shown; and the standardized interpretation reference images are stored according to category, for example: the file L12201041013s1.jpg (Hard floating objects category - Corrugated steel roof houses - January 4, 2022 - Project No. 1 [Wuhan-Guangzhou Railway (Changsha Section, Hunan)] - Safety Hazard No. 013 - Satellite Image - Image No. 1) is stored in the L (Major category: Hard floating objects) folder and the 1 (Minor category: Corrugated steel roof houses) folder.

[0079] Add a record to the Project Information Management table of the database with the following information: "Project Code: 1, Project Name: Wuhan-Guangzhou Railway (Changsha Section, Hunan)"; add a record to the Image Information Management table with the following information: "Code: L12201041013s1, Project Code: 1, Major Safety Hazard Category: Hard Floating Objects, Minor Safety Hazard Category: Corrugated Steel Roof, Longitude: 113.069913, Latitude: 28.292259, Altitude: 26.37, Data Source Category: s, Storage Location: D: / data / L / L1, Geographic Location: Shuiduping, Xianglong Street, Changsha County, Changsha City, Hunan Province, Remarks: ".

[0080] S5, Remote sensing identification of safety hazards in the external environment of railways:

[0081] Satellite imagery of the Wuhan-Guangzhou Railway (Changsha section, Hunan Province), taken by the GF-2 satellite with a high spatial resolution of 1m on [Date] 2021, was used for remote sensing identification of external environmental safety hazards. Image preprocessing included radiometric correction, image fusion, orthorectification, and mosaicking. Deep learning technology was used to initially extract vector patches of safety hazards. The preliminary extraction results were overlaid on the satellite remote sensing imagery. Interpreters then used known safety hazard feature images from the remote sensing interpretation library to identify, interpret, and edit the images based on the preliminary extraction results. After completing the remote sensing extraction of railway external environmental safety hazards, on [Date] 2021, an on-site verification was conducted at the project location. All monitoring results were verified, investigated, recorded, and photographed as evidence.

[0082] S6, Remote Sensing Interpretation Library Update:

[0083] Taking the remote sensing identification project of external environmental safety hazards of the Wuhan-Guangzhou Railway (Changsha section, Hunan) as an example, after completing the remote sensing extraction and on-site verification of safety hazards, the on-site photos of safety hazards are archived and stored in the database. For representative railway external environmental safety hazards, such as the Wuhan-Guangzhou Railway (Changsha section, Hunan) - light floating objects - rainproof cloth - safety hazard No. 001, the on-site photos, satellite images and aerial photography data are interpreted and reference images are created according to step S2. The interpretation library images are encoded according to step S1, namely: M52112291001p1.jpg, M52112291001s1.jpg, M52112291001a1.jpg. The interpretation reference images are stored and the attribute data is stored in the database according to step S4.

Claims

1. A method for identifying safety hazards in the external environment of railways based on a remote sensing interpretation library, comprising the following steps: S1, Remote sensing interpretation library image coding: Develop coding rules for interpretation library images based on the railway external environment safety hazard classification system; S2, Create interpretation reference images: Taking a single safety hazard feature as the object, create on-site photographs, satellite images, aerial photographs and structural diagrams of the object. Each interpretation reference image is uniformly named according to the coding rules of step S1. S3, Establish an interpretation reference image database: Select a relational database to manage the relevant information and project information of the interpretation reference images produced in S2, including an image information management table and a project information management table, and establish a connection relationship between the image information management table and the project information management table through project coding; S4, Interpretation Reference Image Data Management: Based on the interpretation reference image database established in step S3, with the interpretation database image data as the core, the interpretation reference images produced in S2 are classified and stored in JPG format, and the railway external environment safety hazard monitoring project and interpretation reference image attribute information are managed in the form of a database. S5, Remote sensing identification of safety hazards in the external environment of railways: High spatial resolution remote sensing images with a resolution better than 1m are used as the data source for extraction. The extraction range is within 1km on both sides of the railway mainline. Aerial photography image processing and satellite image preprocessing are carried out. Computer automatic extraction and manual interpretation and editing are used to remotely extract safety hazards in the external environment of railways. When performing manual interpretation, the interpretation of safety hazard categories refers to the data resources in the remote sensing interpretation library of safety hazards in the external environment of railways. Finally, a vector image of safety hazards in the external environment of railways is obtained. S6, Remote Sensing Interpretation Library Update: After the completion of the new phase of remote sensing extraction of railway external environmental safety hazards, and in conjunction with field verification data, representative railway external environmental safety hazards are interpreted as follows: reference images are created according to step S2, interpretation library images are encoded according to step S1, and interpretation reference image data is stored in the library according to step S4, thus completing the expansion and correction of the railway external environmental safety hazard remote sensing interpretation library.

2. The method for identifying railway external environmental safety hazards based on a remote sensing interpretation library according to claim 1, characterized in that: The remote sensing interpretation library image encoding in step S1 includes major category of security hazard, minor category of security hazard, acquisition time, project number, risk source number, data source type, and image sequence number.

3. The method for identifying railway external environmental safety hazards based on a remote sensing interpretation library according to claim 1, characterized in that: In step S2, for on-site photographs, the longitude, latitude, and altitude information of the shooting time and location are recorded; for satellite imagery, the safety hazard patches are centered within the output area, the satellite imagery scaling is adjusted to ensure the safety hazard patches are fully displayed without pixelation, a scale bar icon is introduced, and the satellite imagery acquisition time and the longitude and latitude information of the safety hazard center are recorded; for aerial photography, the production process is the same as that for satellite imagery production; for structural diagrams, they are converted to JPG format.

4. The method for identifying railway external environmental safety hazards based on a remote sensing interpretation library according to claim 1, characterized in that: The image information management table mentioned in step S3 includes 11 fields: code, project code, major category of safety hazard, minor category of safety hazard, longitude, latitude, altitude, data source category, storage location, geographical location, and remarks; the project information management table includes 2 fields: project code and project name.

5. The method for identifying railway external environmental safety hazards based on a remote sensing interpretation library according to claim 4, characterized in that: The project coding field serves as a unique identifier to encode the remote sensing identification project for safety hazards in the external environment of the railway, and is associated with the image information management table; The project name field is used to record the naming of projects for remote sensing identification of safety hazards in the external environment of railways. The image information management table is used to record the attribute information of each interpretation reference image, wherein: The encoding field is used to record the decoding library image encoding described in step S1; The project code field is used as a foreign key to connect to the project information management table; The "Safety Hazard Category" field is used to record the category of railway external environment safety hazards to which the image belongs; The safety hazard category field is used to record the railway external environment safety hazard category to which the image belongs; The longitude field is used to record the center longitude coordinates of the associated safety hazard; The latitude field is used to record the center latitude coordinates of the associated safety hazard; The altitude field is used to record the center elevation of the associated safety hazard. The data source category field is used to record the data source type of the interpretation reference image; The storage location field is used to record the file storage path of the interpretation reference image; The geographic location field is used to record the spatial location description of the associated safety hazard; The remarks field is used to record the attribute names of other categories within the subcategory of railway external environmental safety hazards.

Images