CN116071667B - Method and system for detecting abnormal aircraft targets in specified area based on historical data - Google Patents

Method and system for detecting abnormal aircraft targets in specified area based on historical data Download PDF

Info

Publication number
CN116071667B
CN116071667B CN202310361972.2A CN202310361972A CN116071667B CN 116071667 B CN116071667 B CN 116071667B CN 202310361972 A CN202310361972 A CN 202310361972A CN 116071667 B CN116071667 B CN 116071667B
Authority
CN
China
Prior art keywords
data
aircraft
target
representing
flight
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202310361972.2A
Other languages
Chinese (zh)
Other versions
CN116071667A (en
Inventor
周治国
曹宇鹏
周学华
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN202310361972.2A priority Critical patent/CN116071667B/en
Publication of CN116071667A publication Critical patent/CN116071667A/en
Application granted granted Critical
Publication of CN116071667B publication Critical patent/CN116071667B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40—Business processes related to the transportation industry
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
    • G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G5/00—Traffic control systems for aircraft, e.g. air-traffic control [ATC]
    • G08G5/0095—Aspects of air-traffic control not provided for in the other subgroups of this main group
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Software Systems (AREA)
  • Strategic Management (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Medical Informatics (AREA)
  • Databases & Information Systems (AREA)
  • Marketing (AREA)
  • Computational Linguistics (AREA)
  • Tourism & Hospitality (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Development Economics (AREA)
  • Molecular Biology (AREA)
  • Game Theory and Decision Science (AREA)
  • Data Mining & Analysis (AREA)
  • Biophysics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Biomedical Technology (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Astronomy & Astrophysics (AREA)
  • Electromagnetism (AREA)
  • Aviation & Aerospace Engineering (AREA)

Abstract

The invention provides a method and a system for detecting an abnormal aircraft target in a designated area based on historical data, which are used for acquiring satellite-borne synthetic aperture radar data and satellite-borne optical staring imaging data in the designated area; carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data; detecting the image data by adopting a target detection algorithm to locate and identify an aircraft target, and recording pixel coordinates of the aircraft target; converting the pixel coordinates into flight data of the aircraft target; the flight data includes longitude, latitude, altitude, speed, and heading; in order to improve the accuracy of discrimination, the flight data are predicted and prolonged, the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is calculated, and when the multi-factor bidirectional Hausdorff distance is greater than a set similarity threshold, an abnormal aircraft target is identified.

Description

Method and system for detecting abnormal aircraft targets in specified area based on historical data
Technical Field
The invention relates to the technical field of aerospace, in particular to a method and a system for detecting an abnormal airplane target in a designated area based on historical data.
Background
The aircraft track is generally characterized by high density, high speed, low relative speed between targets, poor distinguishing property and the like, is influenced by factors such as insufficient observation means, limited sensor capacity, uncertainty of surrounding environment and the like, and is difficult to acquire the aircraft track of an aircraft target in a strange area at present, so that the abnormal aircraft target is difficult to identify.
Therefore, it is required to combine the available SAR radar satellite data and the optical staring imaging satellite data with the historical empty pipe data in the designated strange area, and solve the outstanding problem by using the track prediction method, so that the abnormal aircraft target in the designated strange area can be detected, and a good foundation is laid for subsequent satellite resource scheduling, abnormal target tracking and information fusion processing.
Disclosure of Invention
The invention provides a method and a system for detecting an abnormal aircraft target in a designated area based on historical data, which are used for solving the technical problem of judging the abnormal target in a designated strange area.
In order to solve the technical problems, the invention provides a method and a system for detecting an abnormal aircraft target in a specified area based on historical data, which comprises the following steps:
step S1: acquiring satellite-borne synthetic aperture radar data in a designated area, and satellite-borne optical staring imaging data;
step S2: carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data; the method comprises the steps of carrying out a first treatment on the surface of the The method comprises the steps of carrying out a first treatment on the surface of the
Step S3: detecting the image data by adopting a target detection algorithm to locate and identify an aircraft target, and recording pixel coordinates of the aircraft target;
step S4: converting the pixel coordinates into flight data of the aircraft target; the flight data includes longitude, latitude, altitude, speed, and heading;
step S5: in order to improve the accuracy of discrimination, the flight data are predicted and prolonged, the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is calculated, and when the multi-factor bidirectional Hausdorff distance is greater than a set similarity threshold, an abnormal aircraft target is identified.
Preferably, the target detection algorithm in step S3 uses a target detection model YOLOv8, and the prior information of the target detection model YOLOv8 includes: the aircraft target is a highlight region and the aircraft shape is shown as a "+" shape.
Preferably, the method of converting the pixel coordinates into flight data of the aircraft target in step S4 includes the steps of:
step S41: let O-uv be the pixel coordinate system, O-xy be the image coordinate system, O c -x c y c z c For camera coordinate system, O w -x w y w z w For the world coordinate system, the pixel coordinates (u, v) are converted into world coordinates:
；
wherein f x And f y The lengths of the focal lengths in the x-axis and y-axis directions are shown, u 0 And v 0 Respectively representing the translation distance of a coordinate system, wherein R represents a rotation matrix and T represents a translation matrix;
step S42: calculating the altitude of the aircraft target:
in the formula, h sar Represents the vertical height of the spaceborne SAR radar from the ground, d represents the distance between the spaceborne SAR radar and the airplane, and alpha represents h sar An included angle between d and d;
step S43: calculating the flying speed v of the airplane target:
；
；
in the method, in the process of the invention,representing a semi-normal function, r representing the earth radius, < +.>And->Respectively representing the target of the airplane at t 1 And t 2 The longitude and latitude coordinates of the moment, L, represent the distance of the aircraft target between the two moments.
Preferably, the prediction and extension of the flight data in step S5 by means of an adaptive transfer information entropy diagram convolutional network model comprises the following steps:
step S511: converting the flight data into a flight path set T:
；
；
；
wherein m represents the number of flight paths, P ln N represents the nth track point of the flight path l, n represents the total number of the flight track points;
step S512: original input track X 0 Multiplied by a random initialization weight factor omega,obtaining a linear transformed track X, calculating transfer entropy among variables through a flight track set, and obtaining a transfer entropy matrix T:
；
in the method, in the process of the invention,values representing the entropy of transfer from Y to X, X and Y representing the track, +.>Representing probability distribution value, x corresponding to track points in historical empty management data t And y t Respectively representing longitude and latitude of the aircraft target t moment, < >>，/>，/>Representing conditional entropy;
step S513: converting transfer entropy matrix into：
；
Wherein X is i And X j The ith and j variables of the track, c represents the causality significance threshold, a=a, respectively ij Representing a causal relationship matrix, the matrix being used as an adjacency matrix;
step S514: using m expanded convolution kernels with different receptive fieldsExtracting node characteristics by->Controlling the information quantity ratio transferred by the convolution kernel:
；
wherein h represents an extracted feature, W i Representing the convolution kernel, x is the input time series,representing convolution operations +.>Representing Hadamard product, b i And b j Representing bias items->And->Representing a nonlinear activation function;
step S515: splicing the features obtained by convolution of different expansion convolution kernels:
；
in the method, in the process of the invention,representing a join operation;
step S516: converting an input aircraft track sequence into a feature matrixThe adjacency relation of the nodes in the matrix H is determined by the adjacency matrix A, the nodes of the graph structure are embedded by the adjacency matrix A, the graph characteristic structure is input into the graph convolutional network to extract deep characteristics, and forward transmission updating of the nodes is carried out through the following formula:
；
wherein H is (l+1) Is the output of layer l+1, H (l) Is the input of layer l+1, sigma represents the activation function, D is the degree matrix of the adjacency matrix A, lambda is the trainable parameter, I is the identity matrix, W (l) Is a weight matrix;
step S517: setting the dimension of the last GNN layer as one to obtain a track predicted value P of a one-dimensional vector t+h 。
Preferably, the adaptive transfer information entropy diagram convolution network model evaluates the prediction performance through relative absolute error, root mean square error and correlation coefficient, and performs model parameter optimization through an adaptive optimization algorithm Adam;
the expression of the relative absolute error RAE is:
；
the expression of the root mean square error RMSE is as follows:
；
the expression of the correlation coefficient CORR is as follows:
；
wherein x is i Representing the actual value of the track sequence, p i The predicted value of the track sequence, n, represents the number of predicted points.
Preferably, step S3 further comprises preprocessing the aircraft target to obtain an actual physical length l of the aircraft target 0 Physical width w 0 And physical area S 0 Comprising the following steps:
step S31: semantic segmentation is carried out on the SAR image data to obtain a binary segmentation result image of the airplane target and the background;
step S32: obtaining the number of pixels occupied by the airplane target expressed as the sectional area, the number of pixels between the farthest two points expressed as the length and the number of pixels between the two points expressed as the width vertical based on the binary segmentation result;
step S33: calculating the actual physical length l of the airplane target by the method of the step S4 0 Physical width w 0 And physical area S 0 ；
Preferably, in step S5, the expression for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is:
；
wherein A and B represent tracks, d (TR A ,TR B ) Representing the multi-factor unidirectional Hausdorff distance, P, from track A to track B a And P b The point of the track is indicated and,is the Euclidean distance of the position feature between two points, < >>Euclidean distance, which is the velocity characteristic between two points, < >>Euclidean distance, which is the heading characteristic between two points, < >>Euclidean distance, which is the length characteristic between two points, < >>Euclidean distance, which is the width characteristic between two points, < >>Euclidean distance, which is the area characteristic between two points, < >>For the multifactor distance between two points, +.>、/>、/>、/>、/>、/>Weights of the distance attribute, the speed attribute, the heading attribute, the length attribute, the width attribute and the area attribute are respectively obtained by training self-adaptive to samples, and the value range of each weight is +.>And satisfy->。
Preferably, the step S5 further includes correcting the multi-factor bidirectional Hausdorff distance, including the following steps:
step S521: physical length l of the flying object 0 Physical width w 0 And physical area S 0 Comparing the aircraft model sizes with the sizes of all aircraft models in the historical empty pipe data;
step S522: the correction coefficient P is calculated by the following formula c ：
；
Wherein, I i Representing the aircraft length, w, corresponding to the aircraft model in the historical empty pipe data i Representing the aircraft width s corresponding to the aircraft model in the historical air traffic control data i Represents the area corresponding to the aircraft model in the historical empty pipe data,representing a discrimination ratio threshold;
step S523: based on the correction coefficient P c Correcting the multi-factor bidirectional Hausdorff distance:
；
in the method, in the process of the invention,representing the modified multi-factor bi-directional Hausdorff distance.
The invention also provides a system for detecting the abnormal aircraft target in the specified area based on the historical data, which is characterized in that: the system comprises a data acquisition module, a data association module, a target identification module, a coordinate conversion module, a track prediction module and an abnormal target discrimination module;
the data acquisition module is used for acquiring satellite-borne synthetic aperture radar data in a designated area and satellite-borne optical staring imaging data;
the data association module is used for carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data;
the target recognition module is used for detecting the image data by adopting a target detection algorithm so as to locate and recognize an aircraft target and record pixel coordinates of the aircraft target;
the coordinate conversion module converts the pixel coordinates into flight data of the aircraft target; the flight data includes longitude, latitude, altitude, speed, and heading;
the flight path prediction module is used for predicting and prolonging the flight data; and the flight path prediction module predicts and prolongs the flight data by adopting an adaptive transfer information entropy diagram convolution network model.
The abnormal target judging module is used for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data, and identifying the aircraft as an abnormal aircraft target when the multi-factor bidirectional Hausdorff distance is larger than a set similarity threshold value.
Preferably, the abnormal target judging module further comprises a bidirectional Hausdorff correcting module, and the bidirectional Hausdorff correcting module performs semantic segmentation on the SAR image data to obtain the actual physical length, physical width and physical area of the aircraft target, and compares the actual physical length, physical width and physical area with the sizes of the aircraft models in the historical empty pipe data to calculate a correction coefficient; and correcting the multi-factor bidirectional Hausdorff distance based on the correction coefficient.
The beneficial effects of the invention at least comprise: in order to ensure the authenticity and reliability of the data, the invention adopts the disclosed satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging video data as data sources to process, thereby avoiding the use of data sources with inaccuracy or larger acquisition difficulty; the method has the advantages that the identification data source is improved by predicting and prolonging the flight data, the similarity is identified by considering multiple factors through the multi-factor bidirectional Hausdorff distance, and the identification accuracy of the abnormal aircraft target is improved.
As an additional technical feature, the multi-factor bidirectional Hausdorff distance is corrected by considering three factors of the physical length, the physical width and the physical area of the flying target, so that the judgment accuracy is further enhanced.
Drawings
FIG. 1 is a flow chart of a method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a relative positional relationship between an SAR radar and an aircraft target in accordance with an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of one-way Hausdorff distance according to an embodiment of the present invention;
fig. 4 is a flow chart of a convolutional network model for adaptively delivering a information entropy diagram in accordance with an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is evident that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by a person skilled in the art without any inventive effort, are intended to be within the scope of the present invention, based on the embodiments of the present invention.
As shown in fig. 1, an embodiment of the present invention provides a method for detecting an abnormal aircraft target in a specified area based on historical data, which is characterized in that: the method comprises the following steps:
step S1: and acquiring satellite-borne synthetic aperture radar data in the designated area, and satellite-borne optical staring imaging data.
In order to ensure the authenticity and reliability of data and remove data with inaccurate data sources or larger acquisition difficulty, the embodiment of the invention adopts the disclosed satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging video data.
Step S2: carrying out space-time data association on satellite-borne synthetic aperture radar data and satellite-borne optical staring imaging data to obtain SAR image data;
step S3: detecting the image data by adopting a target detection algorithm to locate and identify the aircraft target and recording the pixel coordinates of the aircraft target;
in the embodiment of the invention, the target detection algorithm adopts a target detection model YOLOv8, and the prior information of the target detection model YOLOv8 comprises: the aircraft target is a highlight region and the aircraft shape is shown as a "+" shape.
In the embodiment of the invention, the actual physical length l of the airplane target is obtained by preprocessing the airplane target 0 Physical width w 0 And physical area S 0 To facilitate the subsequent discrimination, specifically, the method comprises the following steps:
step S31: semantic segmentation is carried out on SAR image data to obtain a binary segmentation result image of an airplane target and a background;
step S32: obtaining the number of pixels occupied by the airplane target expressed as the sectional area, the number of pixels between the farthest two points expressed as the length and the number of pixels between the two points expressed as the width vertical based on the binary segmentation result;
step S33: square through step S4Calculating the actual physical length l of the airplane target by using the method 0 Physical width w 0 And physical area S 0 ；
Step S4: converting the pixel coordinates into flight data of the aircraft target; flight data includes longitude, latitude, altitude, speed, and heading;
the method for converting pixel coordinates into flight data of an aircraft target comprises the following steps:
step S41: let O-uv be the pixel coordinate system, O-xy be the image coordinate system, O c -x c y c z c For camera coordinate system, O w -x w y w z w For the world coordinate system, the pixel coordinates (u, v) are converted into world coordinates:
；
wherein f x And f y The lengths of the focal lengths in the x-axis and y-axis directions are shown, u 0 And v 0 Respectively representing the translation distance of a coordinate system, wherein R represents a rotation matrix and T represents a translation matrix;
step S42: as shown in fig. 2, the flying height of the aircraft can be obtained from the satellite-borne SAR radar data, and the height of the aircraft target is calculated, which is the relative positional relationship between the satellite and the aircraft target:
；
in the formula, h sar Represents the vertical height of the spaceborne SAR radar from the ground, d represents the distance between the spaceborne SAR radar and the airplane, and alpha represents h sar An included angle between d and d;
step S43: calculating the flying speed v of the airplane target:
；
；
in the method, in the process of the invention,representing a semi-normal function, r representing the earth radius, < +.>And->Respectively representing the target of the airplane at t 1 And t 2 The longitude and latitude coordinates of the moment, L, represent the distance of the aircraft target between the two moments.
Step S5: in order to improve the accuracy of discrimination, the flight data is predicted and prolonged, the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is calculated, and when the multi-factor bidirectional Hausdorff distance is greater than a set similarity threshold value, the abnormal aircraft target is identified.
In the embodiment of the present invention, the prediction and extension of flight data is performed by the adaptive transfer information entropy diagram convolutional network model, as shown in fig. 4, including the following steps:
step S511: converting the flight data into a flight path set T:
；
；
；
wherein m represents the number of flight paths, P ln N represents the nth track point of the flight path l, n represents the total number of the flight track points; in the embodiment of the invention, discrete aircraft track points are connected in time sequence to form one flight track of an aircraft, so that the aircraft track in a designated area is obtained, wherein the initial aircraft target is a short trackThe sequence is only a part of a complete flight path, and the effect of directly judging the abnormal aircraft targets is not good. In order to solve the problem, the short tracks with the track points less than 50 are predicted and prolonged through the subsequent steps, so that the initial short tracks are increased by a certain length, and the judgment of abnormal aircraft targets is facilitated.
Step S512: original input track X 0 Multiplying the linear transformation by a random initialization weight factor omega to obtain a linear transformed track X, and calculating transfer entropy among variables through a flight track set to obtain a transfer entropy matrix T:
；
the transfer entropy can indicate the direction of the information flow, and thus characterize the causal relationship, where,values representing the entropy of transfer from Y to X, X and Y representing the track, +.>Representing probability distribution value, x corresponding to track points in historical empty management data t And y t Respectively representing longitude and latitude of the aircraft target t moment, < >>，/>，Representing conditional entropy;
step S513: converting transfer entropy matrix into：
；
If it isIf the value is greater than zero, the reason that X is Y is described, otherwise Y is the reason of X, wherein X i And X j The ith and j variables of the track, c represents the causality significance threshold, a=a, respectively ij Representing a causal relationship matrix, the matrix being used as an adjacency matrix;
step S514: using m expanded convolution kernels with different receptive fieldsExtracting node characteristics by->Controlling the information quantity ratio transferred by the convolution kernel:
；
wherein h represents an extracted feature, W i Representing the convolution kernel, x is the input time series,representing convolution operations +.>Representing Hadamard product, b i 、b j Representing bias items->、/>Representing a nonlinear activation function.
Step S515: splicing the features obtained by convolution of different expansion convolution kernels:
；
in the method, in the process of the invention,representing a join operation;
step S516: converting an input aircraft track sequence into a feature matrixThe adjacency relation of the nodes in the matrix H is determined by an adjacency matrix A, the nodes of the graph structure are embedded by the adjacency matrix A, the graph characteristic structure is input into the graph convolution network to extract deep characteristics, forward transmission updating of the nodes is carried out by the following formula,
；
wherein H is (l+1) Is the output of layer l+1, H (l) Is the input of layer l+1, sigma represents the activation function, D is the degree matrix of the adjacency matrix A, lambda is the trainable parameter, I is the identity matrix, W (l) Is a weight matrix;
step S517: setting the dimension of the last GNN layer as one to obtain a track predicted value P of a one-dimensional vector t+h 。
In the embodiment of the invention, the adaptive transfer information entropy diagram convolution network model evaluates the prediction performance through relative absolute error, root mean square error and correlation coefficient, and performs model parameter optimization through an adaptive optimization algorithm Adam;
the expression of the relative absolute error RAE is:
；
the expression of the root mean square error RMSE is:
；
the expression of the correlation coefficient CORR is:
；
wherein x is i Representing the actual value of the track, p i Predicted value of track, n represents number of predicted points, p i From track prediction P t+h And intercepting the data.
After the prediction is prolonged, the expression for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is as follows:
；
wherein A and B represent tracks, d (TR A ,TR B ) The multi-factor one-way Hausdorff distance from track A to track B is shown, the one-way Hausdorff distance is shown in figure 3, and P a And P b The point of the track is indicated and,is the Euclidean distance of the position feature between two points, < >>Euclidean distance, which is the velocity characteristic between two points, < >>Euclidean distance, which is the heading characteristic between two points, < >>Euclidean distance, which is the length characteristic between two points, < >>Euclidean distance, which is the width characteristic between two points, < >>Euclidean distance, which is the area characteristic between two points, < >>For the multifactor distance between two points, +.>、/>、/>、/>、/>、/>Weights of the distance attribute, the speed attribute, the heading attribute, the length attribute, the width attribute and the area attribute are respectively obtained by training self-adaptive to samples, and the value range of each weight is +.>And satisfy->。
The embodiment of the invention improves the accuracy of discrimination by correcting the multi-factor bidirectional Hausdorff distance, and specifically comprises the following steps:
step S521: physical length of the object to be flown/ 0 Physical width w 0 And physical area S 0 Comparing the aircraft model sizes with the sizes of all aircraft models in the historical empty pipe data;
step S522: the correction coefficient P is calculated by the following formula c ：
；
Wherein, I i Representing the aircraft length, w, corresponding to the aircraft model in the historical empty pipe data i Representing the aircraft width s corresponding to the aircraft model in the historical air traffic control data i Representing the area corresponding to the aircraft model in the historical empty pipe data;representation ofJudging a proportion threshold value;
step S523: based on correction coefficient P c Correcting the multi-factor bidirectional Hausdorff distance:
；
in the method, in the process of the invention,representing the modified multi-factor bi-directional Hausdorff distance.
The invention also provides a system for detecting the abnormal aircraft target in the specified area based on the historical data, which is characterized in that: the system comprises a data acquisition module, a data association module, a target identification module, a coordinate conversion module, a track prediction module and an abnormal target discrimination module;
the data acquisition module is used for acquiring satellite-borne synthetic aperture radar data in the designated area and satellite-borne optical staring imaging data;
the data association module is used for carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data;
the target recognition module is used for detecting the image data by adopting a target detection algorithm so as to position and recognize the aircraft target and record the pixel coordinates of the aircraft target;
the coordinate conversion module converts the pixel coordinates into flight data of an airplane target; flight data includes longitude, latitude, altitude, speed, and heading;
the flight path prediction module is used for predicting and prolonging flight data;
the abnormal target judging module is used for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data, and identifying the aircraft as an abnormal aircraft target when the multi-factor bidirectional Hausdorff distance is larger than a set similarity threshold value.
And the flight path prediction module predicts and prolongs the flight data by adopting an adaptive transmission information entropy diagram convolution network model.
The abnormal target judging module further comprises a bidirectional Hausdorff correcting module, the bidirectional Hausdorff correcting module performs semantic segmentation on SAR image data to obtain the actual physical length, physical width and physical area of the aircraft target, and the actual physical length, physical width and physical area are compared with the sizes of the aircraft models in the historical empty pipe data to calculate correction coefficients; and correcting the multi-factor bidirectional Hausdorff distance based on the correction coefficient.
The foregoing embodiments may be combined in any way, and all possible combinations of the features of the foregoing embodiments are not described for brevity, but only the preferred embodiments of the invention are described in detail, which should not be construed as limiting the scope of the invention. The scope of the present specification should be considered as long as there is no contradiction between the combinations of these technical features.
It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.

Claims (8)

1. A method for detecting an abnormal aircraft target in a designated area based on historical data is characterized by comprising the following steps: the method comprises the following steps:
step S1: acquiring satellite-borne synthetic aperture radar data and satellite-borne optical staring imaging data in a designated area;
step S2: carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data;
step S3: detecting the image data by adopting a target detection algorithm to locate and identify an aircraft target, and recording pixel coordinates of the aircraft target;
step S4: converting the pixel coordinates into flight data of the aircraft target; the flight data includes longitude, latitude, altitude, speed, and heading;
step S5: in order to improve the accuracy of discrimination, predicting and prolonging the flight data, calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data, and identifying an abnormal aircraft target when the multi-factor bidirectional Hausdorff distance is greater than a set similarity threshold;
in step S5, the flight data is predicted and prolonged by an adaptive information-transferring entropy diagram convolutional network model, which includes the following steps:
step S511: converting the flight data into a flight path set TR:
TR＝{TR 1 ，TR 2 ，…，TR m }
TR l ＝{P l1 ，P l2 ，…，P ln }
P ln = { time, longitude, latitude, altitude, speed, heading }
Wherein m represents the number of flight paths, P ln Representing flight path TR l N represents the total number of flight path points;
step S512: multiplying the original input flight path set TR by a random initialization weight factor omega to obtain a linearly transformed flight path set TR', and calculating transfer entropy among variables through the transformed flight path set to obtain a transfer entropy matrix T m ：
Wherein T is Y→X A value representing the transfer entropy from Y to X, p (·) represents the probability distribution value corresponding to the track points in the historical empty pipe data, X t And y t Representing the longitude and latitude respectively of the aircraft target at time t,H(X t+1 |X t ) Representing conditional entropy;
step S513: converting transfer entropy matrix into T X，Y ：
T X，Y ＝T X→Y -T Y→X
Wherein X is i And X j X i and j values, c represents a causal relationship significance threshold, a=a ij Representing a causal relationship matrix, the matrix being used as an adjacency matrix;
step S514: using q dilation convolution kernels (1 xk) with different receptive fields con ) Con=1, 2, …, q, extract node features:
in the formula, h q Representing the extracted features, W q Represents the q-th dilation convolution kernel, G is the input time series, represents the convolution operation,representing Hadamard product, b 1 And b 2 Representing bias terms, reLU (·) and tanh (·) representing nonlinear activation functions;
step S515: splicing the features obtained by convolution of different convolution kernels:
in the method, in the process of the invention,representing a join operation;
step S516: converting the input flight path set into a feature matrix H E R n×d The adjacency relation of the nodes in the matrix H is determined by the adjacency matrix A, the nodes of the graph structure are embedded by the adjacency matrix A, the graph characteristic structure is input into the graph convolutional network to extract deep characteristics, the forward transmission updating of the nodes is carried out by the following formula,
wherein H is (layer+1) Is the output of layer+1 layer, H (layer) Is the input of layer+1, σ represents the activation function, D is the degree matrix of the adjacency matrix A, λ is the trainable parameter, I is the identity matrix, W (layer) Is a weight matrix;
step S517: setting the dimension of the last GNN layer as one to obtain a track predicted value U of a one-dimensional vector t+s
In step S5, the expression for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data is as follows:
in the formula, TR A And TR B Represents the flight path, d (TR A ，TR B ) Representing flight path TR A To the flight path TR B Multi-factor unidirectional Hausdorff distance, P a And P b Representing the track points, dist (P a ，P b ) Is the euclidean distance of a location feature between two points,euclidean distance, which is the velocity characteristic between two points, < >>Euclidean distance, which is the heading characteristic between two points, < >>Euclidean distance, which is the length characteristic between two points, < >>The euclidean distance of a width feature between two points,multidist (P) a ，P b ) Is the multi-factor distance between two points, w d 、w v 、w θ 、w l 、w w 、w s Weights of the distance attribute, the speed attribute, the heading attribute, the length attribute, the width attribute and the area attribute are respectively obtained by training self-adaptive samples, and the value range of each weight is [0,1]And satisfy w d +w v +w θ +w l +w w +w s ＝1。
2. The method for detecting abnormal aircraft targets in a designated area based on historical data of claim 1, wherein the method comprises the steps of: the target detection algorithm in step S3 adopts a target detection model YOLOv8, and the prior information of the target detection model YOLOv8 includes: the aircraft target is a highlight region and the aircraft shape is shown as a "+" shape.
3. The method for detecting abnormal aircraft targets in a designated area based on historical data of claim 1, wherein the method comprises the steps of: the method for converting the pixel coordinates into flight data of the aircraft target in step S4 comprises the following steps:
step S41: let O-uv be the pixel coordinate system, O-xy be the image coordinate system, O c -x c y c z c For camera coordinate system, O w -x w y w z w For the world coordinate system, the pixel coordinates (u, v) are converted into world coordinates:
wherein f x And f y The lengths of the focal lengths in the x-axis and y-axis directions are shown, u 0 And v 0 Respectively representing the translation distance of a coordinate system, wherein R represents a rotation matrix and T represents a translation matrix;
step S42: calculating altitude h of an aircraft target 0 ：
h 0 ＝h sar -dcosα
In the formula, h sar Represents the vertical height of the spaceborne SAR radar from the ground, d represents the distance between the spaceborne SAR radar and the airplane, and alpha represents h sar An included angle between d and d;
step S43: calculating the flying speed v of the airplane target:
in the method, in the process of the invention,representing a semi-normal function, r representing the earth radius, < +.>And->Respectively representing the target of the airplane at t 1 And t 2 The longitude and latitude coordinates of the moment, L, represent the distance of the aircraft target between the two moments.
4. The method for detecting abnormal aircraft targets in a designated area based on historical data of claim 1, wherein the method comprises the steps of: the adaptive transfer information entropy diagram convolution network model evaluates the prediction performance through relative absolute errors, root mean square errors and correlation coefficients, and performs model parameter optimization through an adaptive optimization algorithm Adam;
the expression of the relative absolute error RAE is:
the expression of the root mean square error RMSE is as follows:
the expression of the correlation coefficient CORR is as follows:
in real in Representing the actual value of the track sequence, p in Representing predicted values of the track sequence, and z represents the number of predicted points.
5. The method for detecting abnormal aircraft targets in a designated area based on historical data of claim 1, wherein the method comprises the steps of: step S3 also comprises preprocessing the airplane target to obtain the actual physical length l of the airplane target 0 Physical width w 0 And physical area S 0 Comprising the following steps:
step S31: semantic segmentation is carried out on the SAR image data to obtain a binary segmentation result image of the airplane target and the background;
step S32: obtaining the number of pixels occupied by the airplane target expressed as the sectional area, the number of pixels between the farthest two points expressed as the length and the number of pixels between the vertical two points expressed as the width based on the binary segmentation result;
step S33: converting the pixel coordinates by the method of the step 4 to calculate the actual physical length l of the aircraft target 0 Physical width w 0 And physical area S 0 。
6. The method for detecting abnormal aircraft targets in a designated area based on historical data of claim 1, wherein the method comprises the steps of: the step S5 also comprises the step of correcting the multi-factor bidirectional Hausdorff distance, and comprises the following steps:
step S521: physical length l of the aircraft target 0 Physical width w 0 And physical area S 0 Comparing the aircraft model sizes with the sizes of all aircraft models in the historical empty pipe data;
step S522: the correction coefficient XZ is calculated by the following formula:
wherein, I i Representing the aircraft length, w, corresponding to the aircraft model in the historical empty pipe data i Representing the aircraft width s corresponding to the aircraft model in the historical air traffic control data i Representing the area corresponding to the aircraft model in the historical empty pipe data; τ represents a discrimination ratio threshold;
step S523: correcting the multi-factor bidirectional Hausdorff distance based on the correction coefficient XZ:
δ(TR A ，TR B ) repair tool ＝δ(TR A ，TR B )×XZ
In the formula, delta (TR A ,TR B ) Repair tool Representing the modified multi-factor bi-directional Hausdorff distance.
7. An abnormal aircraft target detection system in a specified area based on historical data is characterized in that: the method for detecting the abnormal aircraft target in the specified area based on the historical data, which is used for implementing the method for detecting the abnormal aircraft target in the specified area based on the historical data, comprises a data acquisition module, a data association module, a target identification module, a coordinate conversion module, a track prediction module and an abnormal target discrimination module;
the data acquisition module is used for acquiring satellite-borne synthetic aperture radar data in a designated area and satellite-borne optical staring imaging data;
the data association module is used for carrying out space-time data association on the satellite-borne synthetic aperture radar data and the satellite-borne optical staring imaging data to obtain SAR image data;
the target recognition module is used for detecting the image data by adopting a target detection algorithm so as to locate and recognize an aircraft target and record pixel coordinates of the aircraft target;
the coordinate conversion module converts the pixel coordinates into flight data of the aircraft target; the flight data includes longitude, latitude, altitude, speed, and heading;
the flight path prediction module is used for predicting and prolonging the flight data; the flight path prediction module predicts and prolongs the flight data by adopting an adaptive transfer information entropy diagram convolution network model;
the abnormal target judging module is used for calculating the multi-factor bidirectional Hausdorff distance between the flight data and the historical empty pipe data, and identifying the aircraft as an abnormal aircraft target when the multi-factor bidirectional Hausdorff distance is larger than a set similarity threshold value.
8. An aircraft anomaly target detection system in a designated area based on historical data as claimed in claim 7, wherein:
the abnormal target judging module further comprises a multi-factor bidirectional Hausdorff correcting module, the multi-factor bidirectional Hausdorff correcting module performs semantic segmentation on the SAR image data to obtain the actual physical length, physical width and physical area of the aircraft target, and the actual physical length, physical width and physical area are compared with the sizes of the aircraft models in the historical empty pipe data to calculate correction coefficients; and correcting the multi-factor bidirectional Hausdorff distance based on the correction coefficient.
CN202310361972.2A 2023-04-07 2023-04-07 Method and system for detecting abnormal aircraft targets in specified area based on historical data Active CN116071667B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310361972.2A CN116071667B (en) 2023-04-07 2023-04-07 Method and system for detecting abnormal aircraft targets in specified area based on historical data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310361972.2A CN116071667B (en) 2023-04-07 2023-04-07 Method and system for detecting abnormal aircraft targets in specified area based on historical data

Publications (2)

Publication Number Publication Date
CN116071667A CN116071667A (en) 2023-05-05
CN116071667B true CN116071667B (en) 2023-07-18

Family

ID=86170141

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310361972.2A Active CN116071667B (en) 2023-04-07 2023-04-07 Method and system for detecting abnormal aircraft targets in specified area based on historical data

Country Status (1)

Country Link
CN (1) CN116071667B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116229380B (en) * 2023-05-10 2023-09-12 国网江西省电力有限公司电力科学研究院 Method for identifying bird species related to bird-related faults of transformer substation
CN116935700B (en) * 2023-09-18 2023-12-05 四川大学 Sector traffic situation prediction method based on multi-source features

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106022372A (en) * 2016-05-17 2016-10-12 中国人民解放军海军航空工程学院 Target abnormal behavior detection method based multidimensional characteristics

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102902979B (en) * 2012-09-13 2015-08-19 电子科技大学 A kind of method of synthetic-aperture radar automatic target detection
JP2019185179A (en) * 2018-04-03 2019-10-24 株式会社東芝 Target recognition device, target recognition method, and program
CN108596155A (en) * 2018-05-14 2018-09-28 浙江大学 A kind of SAR radars airbound target identifying system
CN114399675A (en) * 2021-11-22 2022-04-26 北京航天自动控制研究所 Target detection method and device based on machine vision and laser radar fusion

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106022372A (en) * 2016-05-17 2016-10-12 中国人民解放军海军航空工程学院 Target abnormal behavior detection method based multidimensional characteristics

Also Published As

Publication number Publication date
CN116071667A (en) 2023-05-05

Similar Documents

Publication Publication Date Title
CN116071667B (en) Method and system for detecting abnormal aircraft targets in specified area based on historical data
EP3918428A1 (en) Automatic optimization of machine learning algorithms in the presence of target datasets
CN111144364B (en) Twin network target tracking method based on channel attention updating mechanism
CN111161315A (en) Multi-target tracking method and system based on graph neural network
JP6892606B2 (en) Positioning device, position identification method and computer program
CN114090718B (en) Bi-LSTM prediction and fuzzy analysis based interrupted track correlation method
Tsintotas et al. Tracking‐DOSeqSLAM: A dynamic sequence‐based visual place recognition paradigm
CN112883850A (en) Multi-view aerospace remote sensing image matching method based on convolutional neural network
CN111046756A (en) Convolutional neural network detection method for high-resolution remote sensing image target scale features
CN114565668A (en) Instant positioning and mapping method and device
CN114608568B (en) Multi-sensor information based instant fusion positioning method
CN116188893A (en) Image detection model training and target detection method and device based on BEV
Khatri et al. Detection of animals in thermal imagery for surveillance using GAN and object detection framework
Kant et al. Long short-term memory auto-encoder-based position prediction model for fixed-wing uav during communication failure
CN113989612A (en) Remote sensing image target detection method based on attention and generation countermeasure network
Penteliuc et al. Prediction of cloud movement from satellite images using neural networks
Braga et al. Lidar and non-extensive particle filter for UAV autonomous navigation
CN113724325B (en) Multi-scene monocular camera pose regression method based on graph convolution network
JP3208634B2 (en) Same target determination device and same target determination method
CN111860613B (en) Multi-target tracking and state predicting method based on multi-source heterogeneous signals
CN114022516A (en) Bimodal visual tracking method based on high rank characteristics and position attention
KR20230023227A (en) Method and Apparatus for Effective Training for Deep Learning-based Precipitation Nowcasting and Estimation
Ying et al. GPK-YOLOv5s: Content-Aware Reassembly of Features and Self Attention for High Altitude Parabolic Detection and Tracking
Zhai et al. Online few-shot learning for multi-function radars mode recognition based on backtracking contextual prototypical memory
CN114693988B (en) Satellite autonomous pose judging method, system and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant