Airport scene information autonomous sensing method based on multi-camera cooperative scheduling

Abstract

The invention discloses an airport scene information autonomous sensing method based on multi-camera cooperative scheduling, and belongs to the technical field of image processing. The method comprises the following steps: collecting image data including the activities of an aircraft and other operation units on a scene, carrying out manual labeling, and training a target detection model; designing an aircraft working state judgment rule, and judging the working state of the scene aircraft; according to ADS-B monitoring data and the working state of the scene airplane, the distribution cost between each airplane and each camera is calculated, a bipartite graph matching theory-based distribution cost matrix is constructed, and a current scheduling scheme of the camera is generated based on the distribution cost matrix, so that the current working state, position coordinates, speed and other operation information of the scene airplane are determined based on the current scheduling scheme; and autonomous perception of airport scene information is realized. Under the condition that the scene information storage cost is greatly reduced, autonomous perception of the scene information is realized through cooperative use of multi-source and multi-mode information, and the intelligent operation and control capability of the airport scene is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of machine learning technology, and in particular to a method for autonomously perceiving airport scene information with collaborative scheduling of multiple cameras. Background Art

[0002] In recent years, my country's civil aviation industry has steadily developed, becoming a multi-sector civil aviation powerhouse. The industry, in line with the requirements of high-quality development, is continuously advancing the development of smart civil aviation. Intelligent airport surface operation control is a key component of smart civil aviation, effectively improving surface operation safety and efficiency. The large number of aircraft, maintenance vehicles, and personnel involved in airport operations, along with the tasks they perform, constitutes the bulk of complex airport surface information. Therefore, efficiently sensing this information is fundamental to achieving intelligent airport surface operation control.

[0003] Airport surfaces contain a vast amount of complex operational information. ADS-B data provides real-time information on aircraft flight information, coordinates, altitude, speed, and other parameters. Optical imaging devices such as cameras provide visual surveillance of specific areas of the surface, capturing image data of the corresponding scene. Furthermore, operational information includes a variety of other data, including pavement conditions and weather information. This information is characterized by rapid updates and diverse modalities, making it difficult for existing sensing methods to accurately and accurately perceive this surface data in real time.

[0004] Existing airport surface information perception methods typically employ a fixed model, indiscriminately, unselectively, and without focus, collecting a wide range of operational information, including flight dynamics, vehicle locations, and equipment status. However, in actual operations, airport traffic levels fluctuate significantly, and the locations and operational requirements of various work units also fluctuate frequently. In this dynamic environment, the continued use of a unified information perception strategy would not only result in the acquisition of large amounts of low-value or duplicate data, but also severely waste monitoring and storage resources, compromising the efficiency of real-time processing of critical operational information. Summary of the Invention

[0005] In order to solve the technical problems existing in the above-mentioned prior art, the present invention provides an autonomous perception method of airport scene information with collaborative scheduling of multiple cameras, aiming to solve the problems of low-value and repeated data acquisition and monitoring and waste of storage resources in existing information perception strategies.

[0006] The technical solution adopted in the present invention is as follows:

[0007] Step 1: Input a number of image data of airport surface activities, including aircraft and other operational objects. Manually label the image data with the categories of the target facilities (e.g., marking the categories of aircraft, various maintenance facilities, and personnel), to obtain the original image dataset.

[0008] Step 2: Perform network training on the constructed target detection network based on the original image dataset to obtain a target detection model;

[0009] Among them, the target detection network can be built based on networks such as YOLOv10, NanoDet or DETR;

[0010] Step 3: Based on the set aircraft operating status determination rules, the target detection model is used to perform target detection on the airport image data collected by the airport scene camera. The aircraft operating status is determined based on the obtained target detection box boundaries and the speed information in the ADS-B surveillance data (Automatic Dependent Surveillance-Broadcast) of the airport scene.

[0011] Step 4: Construct a camera allocation cost matrix based on bipartite graph matching. This matrix is determined by using the coordinate information in the ADS-B surveillance data, the aircraft operating status determination results, and the target detection results of the airport scene camera images output by the target detection model.

[0012] Step 5: Generate a scheduling plan for airport surface cameras based on the allocation cost matrix, and determine the operation information of airport surface aircraft based on the scheduling plan and the current aircraft working status, thereby achieving autonomous perception of airport surface operation information.

[0013] Furthermore, in step 2, when the constructed target detection network is trained based on the original image dataset, the image preprocessing of the original image dataset includes:

[0014] The original image dataset is randomly divided into a training set and a test set according to a pre-set ratio (used to test the detection performance of the trained object detection model);

[0015] Data enhancement processing is performed on the images in the training set, including image denoising, image rotation, and image brightness adjustment.

[0016] Furthermore, there are three operating states of the aircraft: taxiing state, maintenance state, and idle state.

[0017] Furthermore, the determination rules for the three aircraft working states are specifically set as follows:

[0018] Taxiing state: The speed field in the ADS-B surveillance data is not 0;

[0019] Maintenance status: The aircraft speed is 0, and the object detection model detects the aircraft and the maintenance vehicle in the image captured by the airport scene camera, and the intersection over union (IoU) of the detection bounding boxes of the maintenance vehicle and the aircraft is greater than or equal to the preset first threshold.

[0020] Idle state: The aircraft is stopped and the target detection model detects the target of the image captured by the airport scene camera as follows: no maintenance vehicle is detected, or a maintenance vehicle is detected but the IoU between the detection bounding box of the maintenance vehicle and the detected aircraft is less than or equal to a preset second threshold.

[0021] Preferably, the first threshold and the second threshold have the same value, and the setting range is 0.6 to 0.8.

[0022] Furthermore, step 4 includes the following steps:

[0023] Step 4.1: Input the latitude and longitude coordinates of the airport surface camera and extract the latitude and longitude coordinate fields from the ADS-B surveillance data of the aircraft at the airport surface;

[0024] Step 4.2: Based on bipartite graph matching theory, all aircraft and cameras in the airport are modeled as a bipartite graph G = (U, V, E), where the vertex set U represents the aircraft set {a1, a2, ..., a m}, the vertex set V represents the camera node set {c1,c2,...,c n}, the edge set E consists of all aircraft-camera pairs, that is, E={(a i ,c j )|a i ∈U,c j ∈V}; m represents the number of aircraft, n represents the number of cameras, a i represents the i-th aircraft, c j represents the jth camera;

[0025] Step 4.3: Assign weights to each edge in edge set E:

[0026] w(a i ,c j )=f dis (a i ,c j )+f sta (a i )+f det (c j ),

[0027] Among them, f dis (a i ,cj ) represents aircraft a i and camera c j The distance between them is calculated based on the haversine formula;

[0028] f sta (a i ) represents the state value determined based on the aircraft's working state: if the aircraft is in idle state, then f sta (a i )=θ, if the aircraft is in taxiing / maintenance state, then f sta (a i )=0; where θ represents a preset cost threshold, and the cost threshold θ is greater than or equal to f dis (a i ,c j )’s value range upper limit;

[0029] f det (c j ) represents the state value based on the target detection result. If an aircraft is detected in the target detection result, then f det (c j )=0; otherwise, f det (c j )=θ;

[0030] Step 4.4: Based on the weights w(a of each edge of the bipartite graph G = (U, V, E) i ,c j ), construct the camera-aircraft allocation cost matrix W.

[0031] Furthermore, in step 4.3, f dis (a i ,c j ) Specifically:

[0032]

[0033] Where R represents the radius of the earth, lat(·) represents the latitude, and lon(·) represents the longitude.

[0034] Furthermore, step 4 also includes: setting the camera-aircraft allocation cost matrix W to be updated every time interval t to ensure that it reflects the current real matching cost according to the dynamic change characteristics of the aircraft and camera states in the airport.

[0035] Furthermore, step 5 includes the following steps:

[0036] Step 5.1: Assign the cost matrix W to the camera-aircraft, and find the weight w(a) of each aircraft (i.e. each row) for different cameras. i ,c j) minimum value; then delete the value greater than the cost threshold θ among the minimum values found to obtain the valid minimum value set M;

[0037] Step 5.2: Based on the elements in the valid minimum value set M, configure the assigned cameras (i.e., working cameras) of the current scheduling plan with respect to the corresponding camera number index in the camera-aircraft allocation cost matrix W, and use them to perform video surveillance of the airport surface; generate a scheduling plan for airport surface cameras based on the labeled assigned cameras and unassigned cameras; and configure the sampling period of the sparse sampling mode for the unassigned cameras;

[0038] Step 5.3: After each sparse sampling mode is triggered, detect the weights w(a) of each aircraft to different cameras in the camera-aircraft allocation cost matrix W. i ,c j ) whether there is an update; and detect the minimum weight w(a i ,c j ) to determine whether the camera index corresponding to the camera has changed. If so, the valid minimum value set M is updated based on the changed camera index, and the scheduling plan is adjusted. This allows for local matching using the updated edge weights. After completing the local allocation, the results are synchronized with the global allocation results. This step is repeated as the allocation cost matrix is updated to ensure that camera resources are always allocated efficiently.

[0039] Furthermore, in step 5, the operation information of the aircraft on the airport surface is determined based on the scheduling plan and the current aircraft operating status, thereby realizing autonomous perception of the airport surface operation information. Specifically, based on the on-site scheduling plan and the current aircraft operating status, the current operating status, position coordinates, speed and other operation information of the aircraft on the surface are determined to realize autonomous perception of the airport surface operation information.

[0040] The technical solution provided by the present invention brings at least the following beneficial effects:

[0041] From the perspective of target detection, this paper adopts a judgment strategy based on the intersection-union ratio of detection bounding boxes to realize autonomous judgment of the working status of aircraft on the airport scene; based on the bipartite graph matching theory, the camera allocation cost is quantified, and then the cameras are autonomously coordinated and dispatched, realizing autonomous perception of effective scene information. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0043] Figure 1A flowchart of a method for autonomously sensing airport scene information with multi-camera collaborative scheduling provided by an embodiment of the present invention;

[0044] Figure 2 Flowchart of data preprocessing in an embodiment of the present invention;

[0045] Figure 3 A flow chart for generating a matching cost matrix in an embodiment of the present invention;

[0046] Figure 4 This is the structure diagram of the camera autonomous scheduling module. DETAILED DESCRIPTION

[0047] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions of the embodiments of the present invention will be described in detail and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described with reference to the drawings are exemplary and are intended to be used to explain the present invention, and should not be understood as limiting the present invention.

[0048] An embodiment of the present invention provides a method for autonomously perceiving airport surface information through multi-camera collaborative scheduling, comprising: collecting and manually annotating image data containing the activities of aircraft and other operating units on the surface; preprocessing the original images; training a target detection model; designing aircraft operating status determination rules to determine the operating status of surface aircraft; calculating the allocation cost between each aircraft and each camera based on ADS-B surveillance data and the operating status of surface aircraft, and constructing an allocation cost matrix based on bipartite graph matching theory; constructing a camera autonomous scheduling module to achieve autonomous scheduling of cameras; and determining operational information such as the current operating status, position coordinates, and speed of surface aircraft based on the autonomously scheduled cameras, thereby achieving autonomous perception of airport surface information. This invention significantly reduces the storage cost of surface information and, through the collaborative use of multi-source and multi-modal information, designs a real-time and efficient autonomous scheduling method for perception resources, achieving autonomous perception of surface information and effectively improving the intelligent operation and control capabilities of airport surfaces.

[0049] In one embodiment, see Figure 1 The embodiment of the present invention provides a method for autonomously sensing airport scene information with multi-camera coordinated scheduling, comprising the following steps:

[0050] Step 1: scene activity image data collection and annotation;

[0051] Step 2: Data is preprocessed and randomly divided to generate training and test datasets respectively;

[0052] Step 3: Train the target detection model;

[0053] Step 4: Design aircraft operating status determination rules using ADS-B data and target detection bounding boxes;

[0054] Step 5, generate the camera-aircraft allocation cost matrix;

[0055] Step 6: Implement real-time camera scheduling based on the allocation cost matrix;

[0056] Step 7: Determine the current working status and speed of the aircraft on the scene and other operational information to achieve autonomous perception of the scene information.

[0057] The specific implementation of the above steps is as follows:

[0058] In step 1: the original airport scene activity image data is collected using image acquisition equipment, and the data is manually annotated to indicate the work unit category c and bounding box in the scene, thereby obtaining the original image dataset.

[0059] In step 2: randomly divide the original dataset into training set and test set according to the given ratio (e.g. 8:2). Then, Figure 2 The preprocessing shown in the figure is as follows. First, the original dataset is resized and normalized to obtain an image of size 1920*1080. Then, image denoising is performed based on the NLM (Non-Localmeans) algorithm. After that, data augmentation is performed by randomly flipping the image with a probability p and randomly adjusting the image brightness using a normal distribution.

[0060] In step 3, we constructed an object detection model using the YOLOv10-M architecture based on the preprocessed airport scene dataset. We initialized the network parameters through transfer learning and then performed end-to-end training using the Adam optimizer. It should be noted that other object detection models, such as those in the YOLO family, can also be used.

[0061] In step 4: the operating status of the aircraft is determined using the target detection bounding box and the ADS-B message information. In an embodiment of the present invention, the operating status of the aircraft is divided into three types: taxiing status, maintenance status, and idle status. The above three working status determination methods are: taxiing status: the speed field in the ADS-B message broadcast by the aircraft is not 0; maintenance status: the aircraft speed is 0, the camera detects the aircraft and the maintenance vehicle, and the intersection over union (IoU) of the detection bounding box of the maintenance vehicle and the aircraft is greater than the preset threshold. In the present invention, the threshold setting range is 0.6 to 0.8; idle status: the aircraft stops and the camera fails to detect the maintenance vehicle, or the camera can detect the maintenance vehicle but the detection bounding box IoU of the aircraft and the maintenance vehicle is less than the set threshold.

[0062] In step 5: Constructing the camera allocation cost matrix based on bipartite graph matching specifically includes:

[0063] Step 5.1: Collect the latitude and longitude coordinates of all cameras in the scene, and extract the latitude and longitude coordinate information from the ADS-B messages of the aircraft in the scene. Then, based on the bipartite graph matching theory, model all aircraft and cameras in the airport scene as a bipartite graph G = (U, V, E), where the vertex set U represents the aircraft set {a1, a2, ..., a m}, the vertex set V represents the camera node set {c1,c2,...,c n}, the edge set E consists of all aircraft-camera pairs, that is, E={(a i ,c j )|a i ∈U,c j ∈V}. The subscript of a is the aircraft number, the subscript of c is the camera number, m represents the number of aircraft, and n represents the number of cameras.

[0064] Then calculate the distance between the aircraft and each camera. The specific expression is:

[0065]

[0066] Where: R represents the radius of the earth, lat(·) represents the latitude, lon(·) represents the longitude, a i Indicates the first aircraft, c j In this embodiment of the present invention, in order to make the calculated f dis (a i ,c j ) is satisfied, and the upper limit of its value range does not exceed the set cost threshold θ. When calculating, the value of R is the corresponding value when the earth radius is in kilometers, such as 6371. Of course, the value of the cost threshold θ can also be reasonably set so that θ is greater than or equal to f dis (a i ,c j )'s value range upper limit.

[0067] Afterwards, use f sta (a i ) represents aircraft a i The allocation cost of different states is expressed as follows:

[0068]

[0069] Use f det (c j ) represents a camera c j Whether the aircraft can be detected corresponds to the matching cost, the specific expression is as follows:

[0070]

[0071] Finally, the sum of the above three quantities is used to express the aircraft a i With camera c j The weight of the edge between them, that is, the allocation cost, is expressed as follows:

[0072] w(a i ,c j )=f dis (a i ,c j )+f sta (a i )+f det (c j ),

[0073] Of course, you can also set corresponding weight coefficients for the three calculation amounts, and get the allocation cost w(a i ,c j ).

[0074] Calculate the allocation cost between each camera and each aircraft and construct the camera-aircraft allocation cost matrix W∈R m ×n, where m represents the number of aircraft and n represents the number of cameras. Due to the dynamic changes in the status of aircraft and cameras in the airport, it is set to be checked every 1-2 seconds. m,n Perform dynamic updates.

[0075] In step 6: Use the minimum function to allocate the cost matrix W m,n Find the minimum value of each row and delete the value greater than the cost threshold θ to obtain the valid minimum value set M. The specific expression is as follows:

[0076]

[0077] The threshold θ is a relatively large value. In the embodiment of the present invention, the value of θ is set to 800 or 1000, that is, θ is greater than or equal to f dis (a i ,c j )'s value range upper limit.

[0078] Next, the original allocation matrix column index corresponding to all elements in the valid minimum value set M is calculated, which is the assigned camera number. The assigned camera enters normal operation and performs video surveillance, while the unassigned cameras enter sparse sampling mode, sampling every 2-4 seconds, to achieve global matching.

[0079] Since the matching cost matrix will change in real time with aircraft activities, the dynamically updated allocation cost matrix W m,n For rows with changes, the updated allocation cost is used to perform local re-matching according to the above method, and the results are synchronized to the global matching results to ensure that camera resources are always in an efficient allocation state.

[0080] In step 7: Based on the camera scheduling results, the real-time working status, speed, position and other information of the corresponding aircraft are recorded based on the video stream collected by the activated camera, and ultimately the autonomous perception of active scene information based on the autonomous collaborative scheduling of the cameras is realized.

[0081] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0082] Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the technical features being referred to. Thus, a feature defined as "first," "second," etc., may explicitly or implicitly include at least one of the features.

[0083] Any process or method description in a flowchart or otherwise described in this specification may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention pertain.

[0084] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0085] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0086] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention 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. 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 various embodiments of the present invention, and should all be included in the scope of protection of this application.

Claims

1. A method for autonomous perception of airport scene information with multi-camera coordinated scheduling, characterized in that: The following steps are involved: Step 1: Input a number of image data of airport surface activities, including active targets such as aircraft and other operational objects; manually label the image data with the categories of the active target facilities to obtain an original image dataset; Step 2: Perform network training on the constructed target detection network based on the original image dataset to obtain a target detection model; Step 3: Based on the set aircraft operating status determination rules, the target detection model is used to perform target detection on the airport image data collected by the airport surface camera. The aircraft operating status is determined based on the obtained target detection box boundaries and the speed information in the Automatic Dependent Origination - Broadcast (ADS-B) surveillance data of the airport surface. Step 4: Construct a camera allocation cost matrix based on bipartite graph matching. This matrix is determined by using the coordinate information in the ADS-B surveillance data, the aircraft operating status determination results, and the target detection results of the airport scene camera images output by the target detection model. Step 5: Generate a scheduling plan for airport surface cameras based on the allocation cost matrix, and determine the operation information of airport surface aircraft based on the scheduling plan and the current aircraft working status.

2. The method according to claim 1, wherein In step 2, when the constructed target detection network is trained based on the original image dataset, the image preprocessing of the original image dataset includes: The original image dataset is randomly divided into a training set and a test set according to a pre-set ratio; Data enhancement processing is performed on the images in the training set, including image denoising, image rotation, and image brightness adjustment.

3. The method according to claim 1, wherein There are three operating states of an aircraft: taxiing state, maintenance state, and idle state.

4. The method according to claim 3, wherein The specific determination rules for the three aircraft working states are as follows: Taxiing state: The speed field in the ADS-B surveillance data is not 0; Maintenance status: The aircraft speed is 0, and the target detection model detects the aircraft and the maintenance vehicle in the image captured by the airport scene camera, and the intersection over union (IoU) of the detection bounding boxes of the maintenance vehicle and the aircraft is greater than or equal to the preset first threshold. Idle state: The aircraft is stopped and the target detection model detects the target of the image captured by the airport scene camera as follows: no maintenance vehicle is detected, or a maintenance vehicle is detected but the IoU between the detection bounding box of the maintenance vehicle and the detected aircraft is less than a preset second threshold.

5. The method according to claim 4, wherein The first threshold and the second threshold have the same value, and the setting range is 0.6 to 0.

8.

6. The method according to claim 1, wherein Step 4 includes the following steps: Step 4.1: Input the latitude and longitude coordinates of the airport surface camera and extract the latitude and longitude coordinate fields from the ADS-B surveillance data of the airport surface; Step 4.2: Based on bipartite graph matching theory, all aircraft and cameras in the airport are modeled as a bipartite graph G = (U, V, E), where the vertex set U represents the aircraft set {a1, a2, ..., a m }, the vertex set V represents the camera node set {c1,c2,...,c n }, the edge set E consists of all aircraft-camera pairs, that is, E={(a i ,c j )|a i ∈U,c j ∈V}; m represents the number of aircraft, n represents the number of cameras, a i represents the i-th aircraft, c j represents the jth camera; Step 4.3: Assign weights to each edge in edge set E: Among them, f dis (a i ,c j ) represents aircraft a i and camera c j The distance between them is calculated based on the haversine formula; f sta (a i ) represents the state value determined based on the aircraft's working state: if the aircraft is in idle state, then f sta (a i )=θ, if the aircraft is in taxiing / maintenance state, then f sta (a i )=0; where θ represents a preset cost threshold, and the cost threshold θ is greater than or equal to f dis (a i ,c j )’s value range upper limit; f det (c j ) represents the state value based on the target detection result. If an aircraft is detected in the target detection result, then f det (c j )=0; otherwise, f det (c j )=θ; Step 4.4: Based on the weights w(a of each edge of the bipartite graph G = (U, V, E) i ,c j ), construct the camera-aircraft allocation cost matrix W.

7. The method according to claim 8, wherein In step 4.3, f dis (a i ,c j ) Specifically: Where R represents the radius of the earth, lat(·) represents the latitude, and lon(·) represents the longitude.

8. The method according to claim 1, wherein Step 4 also includes: setting the camera-aircraft allocation cost matrix W to be updated every time interval t.

9. The method according to claim 1, wherein Step 5 includes the following steps: Step 5.1: Assign the cost matrix W to the camera-aircraft, and find the weight w(a) of each aircraft to different cameras. i ,c j ) minimum value; then delete the value greater than the cost threshold θ from the minimum values found to obtain the effective minimum value set M; where a i represents the i-th aircraft, c j represents the jth camera; Step 5.2: Based on the elements in the valid minimum value set M, configure the assigned cameras of the current scheduling plan with the corresponding camera index in the camera-aircraft allocation cost matrix W, and use them to monitor the airport scene; generate a scheduling plan for airport scene cameras based on the labeled assigned cameras and unassigned cameras; and configure the sampling period of the sparse sampling mode for the unassigned cameras; Step 5.3: After each sparse sampling mode is triggered, detect the weights w(a) of each aircraft to different cameras in the camera-aircraft allocation cost matrix W. i ,c j ) whether there is an update; and detect the minimum weight w(a i ,c j ) is changed. If so, the valid minimum value set M is updated based on the changed camera number index, and then the scheduling plan is adjusted.

10. The method according to claim 1, wherein In step 5, the operation information includes: the current aircraft working status, aircraft position coordinates and speed.

Images