Train rear collision warning method, device and electronic equipment

By installing lidar and cameras at the rear of the train, obtaining three-dimensional point cloud data and frame images, and calculating collision values, the problem of backward collision in the existing technology is solved, and more accurate collision detection and early warning is achieved, improving Traffic safety.

CN118405170BActive Publication Date: 2024-11-26BEIJING URBAN CONSTR INTELLIGENT CONTROL TECH CO LTD
View PDF 10 Cites 0 Cited by

Patent Information

Application Number
CN202410853139.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-28
Publication Date
2024-11-26
Estimated Expiration
2044-06-28

AI Technical Summary

Technical Problem

The existing train anti-collision system is mainly aimed at the detection of front obstacles or front vehicles, and it is difficult to effectively warn and prevent backward collision accidents in the train workshop.

Method used

Install lidar and camera equipment at the rear end of the train. By acquiring three-dimensional point cloud data and frame images, combining time intervals, first and second predicted collision values ​​are calculated, and early warning information is output to realize the detection of train backward collisions and Early warning.

Benefits of technology

Accurate detection and early warning of train backward collisions is achieved, the safety of rail transit is improved, the occurrence of backward collision accidents is avoided, and the safety of drivers and passengers is ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118405170B_ABST
    Figure CN118405170B_ABST
Patent Text Reader

Abstract

The present application provides a train rearward collision warning method, device and electronic equipment, which relate to the field of traffic safety technology. The method includes: when warning for a rearward collision of a train, three-dimensional point cloud data of the rear of the train collected by a laser radar at the same time and a frame image of the rear of the train collected by a camera device can be first obtained; and a first predicted collision value is determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; a second predicted collision value is determined based on the frame image and the previous frame image of the frame image; and warning information is output based on the first predicted collision value and the second predicted collision value. In this way, the first predicted collision value and the second predicted collision value are combined to jointly measure the degree of collision urgency, which can more accurately assess the degree of danger of the collision, provide a more robust warning for the train, and realize rearward collision detection of the train, so that the driver and passengers / staff in the front train can be warned.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of traffic safety technology, and in particular to a train rearward collision warning method, device and electronic equipment. Background Art

[0002] In order to improve the safety and efficiency of rail transit, it is necessary to install relevant train anti-collision systems to ensure that the train can promptly detect dangerous obstacles or vehicles ahead, issue early warnings or corresponding train controls for situations that may lead to accidents, and ensure the safety and stability of the train throughout the entire operation process.

[0003] At present, train collision avoidance systems mostly install sensing equipment, such as lidar, cameras and other high-precision sensor equipment, on the front of the train to enable the train to actively perceive the area in front of the train. When obstacles or the vehicle in front are detected in the area ahead of the track, they are converted into warning information in emergency situations and presented on the display screen in the cab to provide timely warnings to the driver.

[0004] How to detect possible rear-end collisions (rearward collisions) between trains to warn the driver and passengers / staff in the front train is an urgent problem to be solved by those skilled in the art. Summary of the invention

[0005] The present application provides a train rearward collision warning method, device and electronic equipment, which can realize train rearward collision detection, thereby warning the driver and passengers / staff in the preceding train.

[0006] The present application provides a train rear collision warning method, which is applied to a train, wherein a laser radar and a camera device are installed at the rear end of the train, and the method comprises:

[0007] Acquire, at the same time, the three-dimensional point cloud data of the rear of the train collected by the laser radar and the frame image of the rear of the train collected by the camera device;

[0008] Determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and determine a second predicted collision value based on the frame image and a previous frame image of the frame image;

[0009] Based on the first predicted collision value and the second predicted collision value, warning information is output.

[0010] According to a train rearward collision warning method provided by the present application, the first predicted collision value is determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including:

[0011] Based on at least one of the rear direction data, rear left direction data, rear right direction data, top direction data and reflectivity of the train in the three-dimensional point cloud data, the three-dimensional point cloud data is filtered to obtain target three-dimensional point cloud data;

[0012] The first predicted collision value is determined based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

[0013] According to a train rearward collision warning method provided by the present application, the first predicted collision value is determined based on the target three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including:

[0014] Determine a first distance value corresponding to the three-dimensional point cloud data based on the rear direction data of the train in the target three-dimensional point cloud data;

[0015] The first predicted collision value is determined based on the first distance value, the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, and a second distance value corresponding to the previous frame of three-dimensional point cloud data.

[0016] According to a train rearward collision warning method provided by the present application, the determining of a second predicted collision value based on the frame image and a frame image preceding the frame image includes:

[0017] Extracting a target detection frame in the frame image based on a target detection algorithm, and reducing the target detection frame based on a preset size to obtain a reduced target detection frame;

[0018] Performing feature point detection on the reduced target detection frame to obtain a plurality of first feature points corresponding to the frame image;

[0019] The second predicted collision value is determined based on the multiple first feature points and multiple second feature points corresponding to a frame image before the frame image.

[0020] According to a train rearward collision warning method provided by the present application, the determining of the second predicted collision value based on the multiple first feature points and the multiple second feature points corresponding to the previous frame image of the frame image includes:

[0021] Determining the same plurality of target feature points from the plurality of first feature points and the plurality of second feature points;

[0022] Determine the pixel distance between two target feature points in the frame image and the pixel distance between two target feature points in the previous frame image;

[0023] Determine a plurality of ratios based on the pixel distances of the pairwise target feature points in the frame image and the pixel distances in the previous frame image;

[0024] The second predicted collision value is determined based on the multiple ratios and a time interval between the frame image and the previous frame image.

[0025] According to a train rear collision warning method provided by the present application, the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time is obtained, including:

[0026] Performing target detection and target tracking on the historical frame images behind the train collected by the camera device;

[0027] When the target detection result indicates that other objects exist in the historical frame image, and the target tracking result indicates that the other objects are continuously tracked, the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time is obtained.

[0028] According to a train rearward collision warning method provided by the present application, outputting warning information based on the first predicted collision value and the second predicted collision value includes:

[0029] determining a smaller value of the first predicted collision value and the second predicted collision value;

[0030] When the smaller value is less than or equal to the preset collision value threshold, the warning information is output.

[0031] The present application also provides a train rear collision warning device, which is applied to a train, wherein a laser radar and a camera device are installed at the rear end of the train, and the device comprises:

[0032] an acquisition unit, used to acquire, at the same time, the three-dimensional point cloud data of the rear of the train acquired by the laser radar and the frame image of the rear of the train acquired by the camera device;

[0033] A processing unit, configured to determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and to determine a second predicted collision value based on the frame image and a previous frame image of the frame image;

[0034] An output unit is used to output warning information based on the first predicted collision value and the second predicted collision value.

[0035] According to a train rearward collision warning device provided by the present application, the processing unit is used to determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, including:

[0036] Based on at least one of the rear direction data, rear left direction data, rear right direction data, top direction data and reflectivity of the train in the three-dimensional point cloud data, the three-dimensional point cloud data is filtered to obtain target three-dimensional point cloud data;

[0037] The first predicted collision value is determined based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

[0038] According to a train rearward collision warning device provided by the present application, the processing unit is used to determine the first predicted collision value based on the target three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including:

[0039] Determine a first distance value corresponding to the three-dimensional point cloud data based on the rear direction data of the train in the target three-dimensional point cloud data;

[0040] The first predicted collision value is determined based on the first distance value, the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, and a second distance value corresponding to the previous frame of three-dimensional point cloud data.

[0041] According to a train rearward collision warning device provided by the present application, the processing unit is used to determine a second predicted collision value based on the frame image and a frame image before the frame image, including:

[0042] Extracting a target detection frame in the frame image based on a target detection algorithm, and reducing the target detection frame based on a preset size to obtain a reduced target detection frame;

[0043] Performing feature point detection on the reduced target detection frame to obtain a plurality of first feature points corresponding to the frame image;

[0044] The second predicted collision value is determined based on the multiple first feature points and multiple second feature points corresponding to a frame image before the frame image.

[0045] According to a train rear collision warning device provided by the present application, the processing unit is used to determine the second predicted collision value based on the multiple first feature points and the multiple second feature points corresponding to the previous frame image of the frame image, including:

[0046] Determining the same plurality of target feature points from the plurality of first feature points and the plurality of second feature points;

[0047] Determine the pixel distance between two target feature points in the frame image and the pixel distance between two target feature points in the previous frame image;

[0048] Determine a plurality of ratios based on the pixel distances of the pairwise target feature points in the frame image and the pixel distances in the previous frame image;

[0049] The second predicted collision value is determined based on the multiple ratios and a time interval between the frame image and the previous frame image.

[0050] According to a train rear collision warning device provided by the present application, the acquisition unit is used to acquire the three-dimensional point cloud data behind the train collected by the laser radar at the same time, including:

[0051] Performing target detection and target tracking on the historical frame images behind the train collected by the camera device;

[0052] When the target detection result indicates that other objects exist in the historical frame image, and the target tracking result indicates that the other objects are continuously tracked, the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time is obtained.

[0053] According to a train rearward collision warning device provided by the present application, the output unit is used to output warning information based on the first predicted collision value and the second predicted collision value, including:

[0054] determining a smaller value of the first predicted collision value and the second predicted collision value;

[0055] When the smaller value is less than or equal to the preset collision value threshold, the warning information is output.

[0056] The present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the train rear collision warning method as described in any one of the above items is implemented.

[0057] The present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the train rear collision warning method as described in any one of the above items is implemented.

[0058] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the train rear collision warning method as described in any one of the above items.

[0059] The train rear collision warning method, device and electronic device provided in the embodiment of the present application can first obtain the three-dimensional point cloud data of the rear of the train collected by the laser radar and the frame image of the rear of the train collected by the camera device at the same time when warning for the rear collision of the train; and determine the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; determine the second predicted collision value based on the frame image and the previous frame image of the frame image; and then output the warning information based on the first predicted collision value and the second predicted collision value. In this way, the three-dimensional point cloud data and the frame image are used as the basis for collision warning, no longer relying on the mandatory relative speed measurement or distance measurement, and the first predicted collision value and the second predicted collision value are combined to jointly measure the degree of collision urgency, which can more accurately assess the degree of danger of the collision and provide a more robust warning for the train, so that the train rear collision detection can be realized, so that the driver and passengers / staff in the front train can be warned. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the present application or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0061] Figure 1 A flow chart of a train rearward collision warning method provided in an embodiment of the present application;

[0062] Figure 2 A schematic flow chart of a method for determining a first predicted collision value based on three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data provided in an embodiment of the present application;

[0063] Figure 3 A schematic flow chart of a method for determining a second predicted collision value based on a frame image and a frame image preceding the frame image provided in an embodiment of the present application;

[0064] Figure 4 A schematic diagram of the structure of a train rear collision warning device provided in an embodiment of the present application;

[0065] Figure 5 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0067] In the embodiments of the present application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of the present application, the character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0068] The technical solution provided in the embodiment of the present application can be adapted to the train rearward collision warning scenario. In order to improve the safety and efficiency of rail transit, it is necessary to install relevant train anti-collision systems to ensure that the train can detect dangerous obstacles or vehicles ahead in a timely manner, issue warnings or corresponding train controls for situations that may lead to accidents, and ensure the safety and stability of the train during the entire operation process.

[0069] At present, train collision avoidance systems mostly install sensing equipment, such as lidar, cameras and other high-precision sensor equipment, on the front of the train to enable the train to actively perceive the area in front of the train. When obstacles or the vehicle in front are detected in the area ahead of the track, they are converted into warning information in emergency situations and presented on the display screen in the cab to provide timely warnings to the driver.

[0070] In order to realize rear-end train collision detection and thus warn the driver and passengers / staff in the preceding train, an embodiment of the present application provides a rear-end train collision warning method, which is applied to a train. A laser radar and a camera device are installed at the rear end of the train. The laser radar and the camera device are used to sense the environment behind the train, especially when a line signal system or a train-related communication system fails, a rear-end train collision warning can be realized.

[0071] Among them, the laser radar uses a hybrid solid-state laser radar installed at the rear end of the vehicle, which can collect three-dimensional point cloud data in the environment behind the train in real time. The laser radar needs to have a smaller field of view (Field of View) but a larger detection distance. The same camera equipment, such as a camera sensor, is installed at the rear end of the vehicle, which can collect and decode frame images in the environment behind the train in real time. Since the hardware configuration in this application is similar to the general train forward perception system, hardware reuse can be achieved.

[0072] The train rearward collision warning method provided by the present application will be described in detail below through the following specific embodiments. It is understandable that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0073] Figure 1 A flow chart of a train rear collision warning method provided in an embodiment of the present application, which can be executed by software and / or hardware devices and applied to a train, wherein a laser radar and a camera device are installed at the rear end of the train, for example, see Figure 1 As shown, the train rear collision warning method may include:

[0074] S101. Acquire, at the same time, three-dimensional point cloud data of the rear of the train collected by the laser radar and a frame image of the rear of the train collected by the camera device.

[0075] For example, in an embodiment of the present application, when obtaining three-dimensional point cloud data of the rear of a train collected by a lidar, the lidar data of the rear of the train collected by the lidar can be obtained first; the collected lidar data can be preprocessed to obtain three-dimensional point cloud data of the rear of the train.

[0076] For example, in an embodiment of the present application, preprocessing may include overlapping area processing, noise point removal, point cloud filtering, point cloud matching and other processing, which may be specifically set according to actual needs.

[0077] For example, in the embodiment of the present application, when obtaining a frame image of the rear of the train captured by the camera device, the video stream of the rear of the train captured by the camera device can be first obtained, and the captured video stream can be decoded to obtain frame images one by one. For example, in the embodiment of the present application, the time interval between any two adjacent frame images can be recorded as .

[0078] It can be understood that the 3D point cloud data of the rear of the train collected by the lidar and the frame image of the rear of the train collected by the camera device are data at the same time, that is, time-aligned data, ensuring that the time interval between two adjacent 3D point cloud data is , the time interval between two adjacent frames is .

[0079] For example, in the embodiment of the present application, when acquiring the three-dimensional point cloud data behind the train collected by the laser radar, at least two possible implementation methods may be included:

[0080] In one possible implementation, after the train starts running, the laser radar can be controlled to start and collect three-dimensional point cloud data behind the train in real time, thereby obtaining the three-dimensional point cloud data behind the train collected by the laser radar.

[0081] In another possible implementation, after the train starts running, the camera device can be controlled to start first to collect historical frame images behind the train in real time; and target detection and target tracking are performed on the historical frame images; when the target detection result indicates that other objects exist in the historical frame image, and the target tracking result indicates that other objects are continuously tracked, the lidar is controlled to start, thereby obtaining three-dimensional point cloud data behind the train collected by the lidar.

[0082] For example, other objects may be other trains or obstacles, etc., which may be set according to actual needs.

[0083] For example, in an embodiment of the present application, the target detection algorithm may be a target detection algorithm based on traditional methods, such as a deformable part model (DPM) algorithm, etc.; or, a target detection algorithm based on deep learning, such as a YOLO target detection algorithm, etc., which may be specifically set according to actual needs.

[0084] For example, in an embodiment of the present application, the target tracking algorithm may be a multi-target tracking algorithm, such as a multi-target DeepSort algorithm, etc., which may be specifically set according to actual needs.

[0085] It can be seen that, different from the above possible implementation methods, in this possible implementation method, the camera device is first controlled to start, and when the target detection result indicates that there are other objects in the historical frame image, and the target tracking result indicates that other objects are continuously tracked, it means that there is a potential collision behind the train. In this case, the laser radar is controlled to start, and the three-dimensional point cloud data of the rear of the train is collected by the laser radar. This can effectively avoid unnecessary three-dimensional point cloud data collection operations and subsequent train rear collision detection and warning operations through the laser radar when there are no other objects behind the train, thereby saving processing resources to a certain extent.

[0086] For example, in an embodiment of the present application, taking the YOLO target detection algorithm and the multi-target tracking DeepSort algorithm as examples, target detection and tracking are performed on historical frame images. During the detection process, the detection objects appearing at the edge of the historical frame images can be ignored, and only other objects appearing in the central area of ​​the historical frame images can be focused on.

[0087] It should be noted that in the embodiment of the present application, the curvature of the track where the train is located is assumed to be large, and "other objects" can only appear in the central area of ​​the frame image.

[0088] It can be understood that in the embodiment of the present application, if the YOLO target detection algorithm does not detect valid "other objects", it means that there is no potential collision possibility behind the train, and there is no need to perform unnecessary 3D point cloud data acquisition operations and subsequent train rear collision detection and warning operations through the lidar; if the YOLO target detection algorithm detects valid "other objects", but the multi-target tracking DeepSort algorithm does not continuously track the "other objects" in the historical frame images, it means that the tracking is relatively unstable, which may be due to misdetection or because the object is still too far away to be clearly identified, then ... It is necessary to perform unnecessary 3D point cloud data collection operations and subsequent rear train collision detection and warning operations through the lidar; however, if the multi-target tracking DeepSort algorithm is used, "other objects" are continuously tracked in the historical frame images, indicating that there is a potential collision behind the train. In this case, the lidar is controlled to start and the 3D point cloud data of the rear of the train is collected by the lidar. This can effectively avoid the unnecessary 3D point cloud data collection operations and subsequent rear train collision detection and warning operations through the lidar when there are no other objects behind the train, thereby saving processing resources to a certain extent.

[0089] After respectively acquiring the three-dimensional point cloud data of the rear of the train collected by the laser radar and the frame image of the rear of the train collected by the camera device, the following S102 may be executed:

[0090] S102, determining a first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and determining a second predicted collision value based on the frame image and a previous frame image of the frame image.

[0091] Among them, the predicted collision value can be represented by TTC (Time to Collision), which usually refers to the time when two objects or entities are expected to collide while maintaining their current speed and trajectory.

[0092] For example, in the embodiment of the present application, the first predicted collision value can be recorded as , the second predicted collision value can be recorded as , which can be set according to actual needs.

[0093] It should be noted that in the embodiment of the present application, there is no order between determining the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data and determining the second predicted collision value based on the frame image and the previous frame of the frame image. The first predicted collision value can be determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, and then the second predicted collision value can be determined based on the frame image and the previous frame of the frame image; the second predicted collision value can also be determined based on the frame image and the previous frame of the frame image, and then the first predicted collision value can be determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; of course, the first predicted collision value can also be determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, and the second predicted collision value can be determined based on the frame image and the previous frame of the frame image. The specific settings can be made according to actual needs.

[0094] After determining the first predicted collision value based on the 3D point cloud data and the time interval between the 3D point cloud data and the previous frame of 3D point cloud data, and determining the second predicted collision value based on the frame image and the previous frame image of the frame image, the following S103 may be performed:

[0095] S103: Output warning information based on the first predicted collision value and the second predicted collision value.

[0096] The warning information is used to warn the preceding train, that is, the driver and passengers / staff of the train.

[0097] When warning the train driver and passengers / staff, the warning signal includes, on the one hand, a real-time warning of a possible rear-end collision to the train driver; on the other hand, in an emergency (i.e. when a collision is unavoidable), a warning of an impending collision will be sent to all passengers / staff on the train through the on-board broadcasting system, prompting all passengers / staff to hold on firmly and protect themselves to minimize the consequences that may result from the accident.

[0098] It can be seen that in the embodiment of the present application, when warning of a rearward collision of a train, the three-dimensional point cloud data of the rear of the train collected by the laser radar and the frame image of the rear of the train collected by the camera device at the same time can be first obtained; and the first predicted collision value is determined based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; the second predicted collision value is determined based on the frame image and the previous frame image of the frame image; and the warning information is output based on the first predicted collision value and the second predicted collision value. In this way, the three-dimensional point cloud data and the frame image are used as the basis for collision warning, which no longer depends on the mandatory relative speed measurement or distance measurement, and the first predicted collision value and the second predicted collision value are combined to jointly measure the degree of collision urgency, which can more accurately assess the degree of danger of the collision and provide a more robust warning for the train. In this way, the rearward collision detection of the train can be realized, so that the driver and passengers / staff in the front train can be warned.

[0099] Based on the above Figure 1 In order to facilitate understanding of how to determine the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data in the above S102, the following will be described. Figure 2 The illustrated embodiments are described in detail.

[0100] Figure 2 A schematic flow chart of a method for determining a first predicted collision value based on three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data provided in an embodiment of the present application. For example, see Figure 2 As shown, the method may include:

[0101] S201. Filter the three-dimensional point cloud data based on at least one of the data in the rear direction, the left side data, the right side data, the data directly above, and the reflectivity of the train in the three-dimensional point cloud data to obtain target three-dimensional point cloud data.

[0102] For example, in an embodiment of the present application, a vehicle coordinate system can be established based on the position of a laser radar installed at the rear end of the train, where the x direction in the coordinate system is directly behind the train, the y direction in the coordinate system is directly to the left of the rear of the train, and the z direction in the coordinate system is directly above the train.

[0103] For example, in the embodiment of the present application, based on the data of the rear direction of the train in the three-dimensional point cloud data, when filtering the three-dimensional point cloud data, the three-dimensional point cloud data can be filtered. The data in the positive and negative directions are filtered out to obtain the three-dimensional point cloud data after filtering based on the data in the positive and negative directions. Represents the data in the rear direction of the train in the 3D point cloud data.

[0104] For example, in the embodiment of the present application, based on the left data of the rear direction of the train or the right data of the rear direction of the train in the three-dimensional point cloud data, when filtering the three-dimensional point cloud data, the three-dimensional point cloud data can be filtered. or The data is filtered out to obtain the 3D point cloud data after filtering based on the left data in the rear direction. Indicates the left side data or the right side data of the rear direction of the train in the 3D point cloud data. Indicates the size reserved in the y direction. Indicates the width of the track the train is on.

[0105] It can be understood that in the embodiment of the present application, considering the large curvature of the track where the train is located, the curve situation of "other objects" within the effective sensing detection range can be ignored, and usually only the position is: and Therefore, the three-dimensional point cloud data can be or Filter out the data.

[0106] For example, in the embodiment of the present application, based on the data directly above the train in the three-dimensional point cloud data, when filtering the three-dimensional point cloud data, the three-dimensional point cloud data can be filtered. The data is filtered out to obtain the 3D point cloud data after filtering based on the data directly above. Represents the data directly above the train in the 3D point cloud data. Indicates the size reserved in the z direction. Indicates the installation height of the laser radar from the ground.

[0107] For example, in the embodiment of the present application, when filtering the three-dimensional point cloud data based on the reflectivity in the three-dimensional point cloud data, the three-dimensional point cloud data may be: The data is filtered out to obtain the three-dimensional point cloud data after the filtering operation based on reflectivity. Represents the reflectivity of data in 3D point cloud data. Indicates a preset reflectivity threshold.

[0108] By way of example, in an embodiment of the present application, the three-dimensional point cloud data can be filtered together based on the data in the direction directly behind the train, the data in the direction to the left of the train, the data in the direction directly behind the train, the data directly above the train, and the reflectivity in the three-dimensional point cloud data. The specific implementation is similar to the method of filtering the three-dimensional point cloud data separately based on the data in the direction directly behind the train, the data in the direction to the left of the train, the data in the direction directly behind the train, the data directly above the train, and the reflectivity in the three-dimensional point cloud data. Please refer to the above-mentioned relevant description, and it will not be repeated here in the embodiment of the present application. This can minimize the amount of invalid three-dimensional point cloud data, and provide convenience for the subsequent determination of the first predicted collision value based on the target three-dimensional point cloud data.

[0109] S202: Determine a first predicted collision value based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

[0110] By way of example, in an embodiment of the present application, when determining a first predicted collision value based on target three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, a first distance value corresponding to the three-dimensional point cloud data can be first determined based on the rear direction data of the train in the target three-dimensional point cloud data; and the first predicted collision value can be determined based on the first distance value, the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, and the second distance value corresponding to the previous frame of three-dimensional point cloud data.

[0111] By way of example, in an embodiment of the present application, when determining the first distance value corresponding to the current moment based on the rear direction data in the target three-dimensional point cloud data, the median of the rear direction data in the target three-dimensional point cloud data can be determined as the first distance value, that is, the median of the coordinates in the x-direction in the target three-dimensional point cloud data, thereby obtaining the first distance value corresponding to the current moment.

[0112] For example, in the embodiment of the present application, based on the first distance value, the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, and the second distance value corresponding to the previous frame of three-dimensional point cloud data, when determining the first predicted collision value, refer to the following formula 1:

[0113] Formula 1

[0114] in, represents the first predicted collision value, Indicates the first distance value corresponding to the 3D point cloud data, Indicates the time interval between the 3D point cloud data and the previous frame of 3D point cloud data. Indicates the second distance value corresponding to the previous frame of 3D point cloud data.

[0115] It should be noted that in the embodiment of the present application, when determining the second distance value corresponding to the previous frame of three-dimensional point cloud data, its specific implementation is similar to the above-mentioned method of determining the first distance value corresponding to the three-dimensional point cloud data. For details, please refer to the above-mentioned description of determining the first distance value corresponding to the three-dimensional point cloud data. Here, the embodiment of the present application will not be repeated.

[0116] It can be seen that in the embodiment of the present application, when determining the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, the three-dimensional point cloud data can be first filtered based on at least one of the data in the three-dimensional point cloud data in the direction directly behind the train, the data in the left side of the direction directly behind the train, the data in the right side of the direction directly behind the train, the data directly above the train, and the reflectivity to obtain the target three-dimensional point cloud data; and based on the data in the direction directly behind the train in the target three-dimensional point cloud data, determine the first distance value corresponding to the three-dimensional point cloud data. In this way, by filtering the three-dimensional point cloud data, the amount of invalid three-dimensional point cloud data can be effectively reduced, thereby providing convenience for the subsequent determination of the first predicted collision value based on the target three-dimensional point cloud data.

[0117] Based on the above Figure 1 or Figure 2 In order to facilitate understanding of how to determine the second predicted collision value based on the frame image and the previous frame image in the above S102, the following will be described. Figure 3 The illustrated embodiments are described in detail.

[0118] Figure 3 A schematic flow chart of a method for determining a second predicted collision value based on a frame image and a frame image before the frame image provided in an embodiment of the present application. For example, see Figure 3 As shown, the method may include:

[0119] S301 , extracting a target detection frame in a frame image based on a target detection algorithm, and reducing the target detection frame based on a preset size to obtain a reduced target detection frame.

[0120] Among them, the value of the preset size can be set according to actual needs, for example, it can be 0.8 times the target detection frame, or 0.9 times the target detection frame, etc., and can be set according to actual needs.

[0121] For example, in the embodiments of the present application, the target detection algorithm may be a YOLO target detection algorithm. Of course, it may also be other target detection algorithms, which may be set according to actual needs.

[0122] After extracting the target detection frame in the frame image based on the target detection algorithm, considering that the target detection frame may have a size deviation from the actual size occupied by the "target object" in the frame image, the target detection frame can be reduced based on the preset size to obtain a reduced target detection frame. The width of the reduced target detection frame is , the height is .

[0123] After the target detection frame is reduced to obtain the reduced target detection frame, the following S302 can be executed:

[0124] S302: Perform feature point detection on the reduced target detection frame to obtain a plurality of first feature points corresponding to the frame image.

[0125] For example, in an embodiment of the present application, a feature point detection algorithm can be used to perform feature point detection on the reduced target detection frame to obtain multiple first feature points, each of which contains its pixel coordinate information and its descriptor vector, which can be specifically set according to actual needs.

[0126] For example, in an embodiment of the present application, the first feature point may be a Shi-Tomasi feature point, or an oriented FAST and rotated BRIEF (ORB) feature point, etc. It may be set according to actual needs, and the embodiment of the present application does not impose any specific restrictions thereto.

[0127] After determining the multiple first feature points corresponding to the frame image through the feature point detection algorithm, the second predicted collision value can be determined together with the multiple second feature points corresponding to the previous frame image of the frame image, that is, the following S303 is executed:

[0128] S303: Determine a second predicted collision value based on the plurality of first feature points and a plurality of second feature points corresponding to a previous frame image of the frame image.

[0129] By way of example, in an embodiment of the present application, when jointly determining the second predicted collision value based on multiple first feature points and multiple second feature points corresponding to a previous frame image, multiple identical target feature points can be first determined from the multiple first feature points and the multiple second feature points; and the pixel distances between each pair of target feature points in the frame image and the pixel distances in the previous frame image are determined; multiple ratios are determined based on the pixel distances between each pair of target feature points in the frame image and the pixel distances in the previous frame image; and the second predicted collision value is determined based on the multiple ratios and the time interval between the frame image and the previous frame image.

[0130] By way of example, in an embodiment of the present application, when determining multiple ratios based on the pixel distances of pairwise target feature points in the frame image and the pixel distances in the previous frame image, for pairwise target feature points, the ratios of the pixel distances of the pairwise target feature points in the frame image to the pixel distances in the previous frame image can be calculated to obtain the ratios corresponding to the pairwise target feature points, and so on, thereby obtaining multiple ratios.

[0131] For example, in an embodiment of the present application, when determining the second predicted collision value based on multiple ratios and the time interval between a frame image and a previous frame image, a target ratio as a median can be first determined from the multiple ratios, and the second predicted collision value is determined based on the target ratio and the time interval between the frame image and the previous frame image, as shown in the following formula 2:

[0132] Formula 2

[0133] in, represents the second predicted collision value, The time interval between a frame image and the previous frame image, Indicates the target ratio.

[0134] It can be seen that in the embodiment of the present application, when determining the second predicted collision value based on the frame image and the previous frame image of the frame image, the target detection frame in the frame image can be first extracted based on the target detection algorithm, and the target detection frame can be reduced based on a preset size to obtain a reduced target detection frame; and feature point detection is performed on the reduced target detection frame to obtain multiple first feature points; and then based on the multiple first feature points and the multiple second feature points corresponding to the previous frame image of the frame image, the second predicted collision value is determined. In this way, the frame image is used as a basis for collision warning to determine the second predicted collision value, and then the first predicted collision value is combined to jointly measure the degree of collision urgency, so that the degree of collision danger can be more accurately assessed, and a more robust warning can be provided for the train. In this way, rearward collision detection of the train can be realized, so that the driver and passengers / staff in the front train can be warned.

[0135] Based on any of the above embodiments, after determining the first predicted collision value based on the three-dimensional point cloud data and determining the second predicted collision value based on the frame image, warning information can be output based on the first predicted collision value and the second predicted collision value.

[0136] For example, in an embodiment of the present application, based on the first predicted collision value and the second predicted collision value, when outputting warning information, the smaller value of the first predicted collision value and the second predicted collision value can be determined first; and it is determined whether the smaller value is less than or equal to the preset collision value threshold. When the smaller value is greater than the preset collision value threshold, it indicates that a collision will not occur at the rear of the train for the time being; on the contrary, when the smaller value is less than or equal to the preset collision value threshold, it indicates that a collision may occur at the rear of the train, and a warning information can be output, so that the driver and passengers / staff in the front train can be warned.

[0137] The train rear collision warning device provided in the present application is described below. The train rear collision warning device described below and the train rear collision warning method described above can be referenced to each other.

[0138] Figure 4 A schematic diagram of a rear-end train collision warning device provided in an embodiment of the present application is applied to a train, wherein a laser radar and a camera device are installed at the rear end of the train. For example, see Figure 4 As shown, the train rearward collision warning device 40 may include:

[0139] An acquisition unit 401 is used to acquire, at the same time, the three-dimensional point cloud data of the rear of the train acquired by the laser radar and the frame image of the rear of the train acquired by the camera device;

[0140] The processing unit 402 is used to determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and determine a second predicted collision value based on the frame image and a previous frame image of the frame image;

[0141] The output unit 403 is configured to output warning information based on the first predicted collision value and the second predicted collision value.

[0142] For example, in the embodiment of the present application, the processing unit 402 is used to determine the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including:

[0143] Based on at least one of the rear direction data, rear left direction data, rear right direction data, top direction data and reflectivity of the train in the three-dimensional point cloud data, the three-dimensional point cloud data is filtered to obtain target three-dimensional point cloud data;

[0144] The first predicted collision value is determined based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

[0145] For example, in the embodiment of the present application, the processing unit 402 is used to determine the first predicted collision value based on the target three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including:

[0146] Determine a first distance value corresponding to the three-dimensional point cloud data based on the rear direction data of the train in the target three-dimensional point cloud data;

[0147] The first predicted collision value is determined based on the first distance value, the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, and a second distance value corresponding to the previous frame of three-dimensional point cloud data.

[0148] For example, in the embodiment of the present application, the processing unit 402 is used to determine the second predicted collision value based on the frame image and the frame image before the frame image, including:

[0149] Extracting a target detection frame in the frame image based on a target detection algorithm, and reducing the target detection frame based on a preset size to obtain a reduced target detection frame;

[0150] Performing feature point detection on the reduced target detection frame to obtain a plurality of first feature points corresponding to the frame image;

[0151] The second predicted collision value is determined based on the multiple first feature points and multiple second feature points corresponding to a frame image before the frame image.

[0152] For example, in the embodiment of the present application, the processing unit 402 is used to determine the second predicted collision value based on the multiple first feature points and the multiple second feature points corresponding to the previous frame image of the frame image, including:

[0153] Determining the same plurality of target feature points from the plurality of first feature points and the plurality of second feature points;

[0154] Determine the pixel distance between two target feature points in the frame image and the pixel distance between two target feature points in the previous frame image;

[0155] Determine a plurality of ratios based on the pixel distances of the pairwise target feature points in the frame image and the pixel distances in the previous frame image;

[0156] The second predicted collision value is determined based on the multiple ratios and a time interval between the frame image and the previous frame image.

[0157] For example, in the embodiment of the present application, the acquisition unit 401 is used to acquire the three-dimensional point cloud data behind the train collected by the laser radar at the same time, including:

[0158] Performing target detection and target tracking on the historical frame images behind the train collected by the camera device;

[0159] When the target detection result indicates that other objects exist in the historical frame image, and the target tracking result indicates that the other objects are continuously tracked, the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time is obtained.

[0160] For example, in the embodiment of the present application, the output unit 403 is used to output warning information based on the first predicted collision value and the second predicted collision value, including:

[0161] determining a smaller value of the first predicted collision value and the second predicted collision value;

[0162] When the smaller value is less than or equal to the preset collision value threshold, the warning information is output.

[0163] The train rear collision warning device 40 provided in the embodiment of the present application can execute the technical solution of the train rear collision warning method in any of the above-mentioned embodiments. Its implementation principle and beneficial effects are similar to the implementation principle and beneficial effects of the train rear collision warning method. Please refer to the implementation principle and beneficial effects of the train rear collision warning method, which will not be repeated here.

[0164] Figure 5 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present application, such as Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the above-mentioned train rear collision warning method, which includes: obtaining the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time, and the frame image of the rear of the train collected by the camera device; determining a first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; and determining a second predicted collision value based on the frame image and the previous frame image of the frame image; outputting warning information based on the first predicted collision value and the second predicted collision value.

[0165] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0166] On the other hand, the present application also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the above-mentioned train rear collision warning method, which includes: obtaining three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time, and a frame image of the rear of the train collected by the camera device; determining a first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data; and determining a second predicted collision value based on the frame image and the previous frame image of the frame image; outputting warning information based on the first predicted collision value and the second predicted collision value.

[0167] On the other hand, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the above-mentioned train rear collision warning method, the method comprising: obtaining three-dimensional point cloud data of the rear of the train collected by the laser radar and a frame image of the rear of the train collected by the camera device at the same time; determining a first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and determining a second predicted collision value based on the frame image and a previous frame image of the frame image; and outputting warning information based on the first predicted collision value and the second predicted collision value.

[0168] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0169] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A train rearward collision warning method, characterized in that: Applied to a train, the rear end of the train is equipped with a laser radar and a camera device, and the method includes: Acquire, at the same time, the three-dimensional point cloud data of the rear of the train collected by the laser radar and the frame image of the rear of the train collected by the camera device; Determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and determine a second predicted collision value based on the frame image and a previous frame image of the frame image; outputting warning information based on the first predicted collision value and the second predicted collision value; Wherein, determining the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data includes: Based on at least one of the rear direction data, rear left direction data, rear right direction data, top direction data and reflectivity of the train in the three-dimensional point cloud data, the three-dimensional point cloud data is filtered to obtain target three-dimensional point cloud data; The first predicted collision value is determined based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

2. The method according to claim 1, characterized in that The determining the first predicted collision value based on the target three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data includes: Determine a first distance value corresponding to the three-dimensional point cloud data based on the rear direction data of the train in the target three-dimensional point cloud data; The first predicted collision value is determined based on the first distance value, the time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data, and a second distance value corresponding to the previous frame of three-dimensional point cloud data.

3. The method according to claim 1 or 2, characterized in that: The determining a second predicted collision value based on the frame image and a frame image preceding the frame image comprises: Extracting a target detection frame in the frame image based on a target detection algorithm, and reducing the target detection frame based on a preset size to obtain a reduced target detection frame; Performing feature point detection on the reduced target detection frame to obtain a plurality of first feature points corresponding to the frame image; The second predicted collision value is determined based on the multiple first feature points and multiple second feature points corresponding to a frame image before the frame image.

4. The method according to claim 3, characterized in that The determining the second predicted collision value based on the plurality of first feature points and a plurality of second feature points corresponding to a previous frame image of the frame image includes: Determining a plurality of identical target feature points from the plurality of first feature points and the plurality of second feature points; Determine the pixel distance between two target feature points in the frame image and the pixel distance between two target feature points in the previous frame image; Determine a plurality of ratios based on the pixel distances of the pairwise target feature points in the frame image and the pixel distances in the previous frame image; The second predicted collision value is determined based on the multiple ratios and a time interval between the frame image and the previous frame image.

5. The method according to claim 1 or 2, characterized in that: The obtaining of the three-dimensional point cloud data behind the train collected by the laser radar at the same time includes: Performing target detection and target tracking on the historical frame images behind the train collected by the camera device; When the target detection result indicates that other objects exist in the historical frame image, and the target tracking result indicates that the other objects are continuously tracked, the three-dimensional point cloud data of the rear of the train collected by the laser radar at the same time is obtained.

6. The method according to claim 1 or 2, characterized in that: The outputting warning information based on the first predicted collision value and the second predicted collision value includes: determining a smaller value of the first predicted collision value and the second predicted collision value; When the smaller value is less than or equal to the preset collision value threshold, the warning information is output.

7. A train rearward collision warning device, characterized in that: Applied to a train, the rear end of the train is equipped with a laser radar and a camera device, and the device includes: an acquisition unit, used to acquire, at the same time, the three-dimensional point cloud data of the rear of the train acquired by the laser radar and the frame image of the rear of the train acquired by the camera device; A processing unit, configured to determine a first predicted collision value based on the three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data; and to determine a second predicted collision value based on the frame image and a previous frame image of the frame image; an output unit, configured to output warning information based on the first predicted collision value and the second predicted collision value; The processing unit is used to determine the first predicted collision value based on the three-dimensional point cloud data and the time interval between the three-dimensional point cloud data and the previous frame of three-dimensional point cloud data, including: Based on at least one of the rear direction data, rear left direction data, rear right direction data, top direction data and reflectivity of the train in the three-dimensional point cloud data, the three-dimensional point cloud data is filtered to obtain target three-dimensional point cloud data; The first predicted collision value is determined based on the target three-dimensional point cloud data and a time interval between the three-dimensional point cloud data and a previous frame of three-dimensional point cloud data.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the train rear collision warning method as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the train rear collision warning method according to any one of claims 1 to 6 is implemented.

Citation Information

Patent Citations

  • Vehicle speed determination method and vehicle

    CN111986472A

  • Collision early warning system based on fusion of foresight camera and rear millimeter wave radar

    CN114559960A

  • Highway shunting area anti-collision method based on roadside laser radar

    CN115273549A

  • Vehicle low-speed emergency braking method and system, electronic equipment and storage medium

    CN116022134A

  • Barrier detection method, system and device for shunting scene and storage medium

    CN116469071A