CN112329627A - High-altitude throwing object distinguishing method - Google Patents

High-altitude throwing object distinguishing method Download PDF

Info

Publication number
CN112329627A
CN112329627A CN202011224115.0A CN202011224115A CN112329627A CN 112329627 A CN112329627 A CN 112329627A CN 202011224115 A CN202011224115 A CN 202011224115A CN 112329627 A CN112329627 A CN 112329627A
Authority
CN
China
Prior art keywords
throwing object
throwing
wind
iteration
layer
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.)
Granted
Application number
CN202011224115.0A
Other languages
Chinese (zh)
Other versions
CN112329627B (en
Inventor
闫政
杜勇
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.)
Chongqing Lanhui Information Technology Co ltd
Original Assignee
Chongqing Lanhui Information Technology Co ltd
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 Chongqing Lanhui Information Technology Co ltd filed Critical Chongqing Lanhui Information Technology Co ltd
Priority to CN202011224115.0A priority Critical patent/CN112329627B/en
Publication of CN112329627A publication Critical patent/CN112329627A/en
Application granted granted Critical
Publication of CN112329627B publication Critical patent/CN112329627B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
    • G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Computational Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Multimedia (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Algebra (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Operations Research (AREA)
  • Databases & Information Systems (AREA)
  • Molecular Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computing Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Aerodynamic Tests, Hydrodynamic Tests, Wind Tunnels, And Water Tanks (AREA)

Abstract

The invention discloses a high-altitude throwing object distinguishing method, which comprises the following steps of: (S1) the camera photographs the floor of the building; (S2) detecting and extracting a moving object in the camera view using a frame difference method; (S3) extracting a motion trajectory coordinate sequence of the moving object; (S4) inputting the motion trail coordinate sequence of the moving target into a pre-trained neural network discrimination model for classification discrimination, and discriminating whether the moving target belongs to a high altitude parabola. By the high-altitude throwing object distinguishing method, the high-altitude throwing object motion detection and identification are converted into a sequence classification problem, the detection performance is improved, and meanwhile, the complexity of a detection system is reduced.

Description

High-altitude throwing object distinguishing method
Technical Field
The invention relates to the technical field of high-altitude throwing object detection, in particular to a high-altitude throwing object distinguishing method.
Background
The high-altitude parabolic motion is called 'pain over the city', the high-altitude parabolic motion is concerned all the time, and the social hazard brought by the high-altitude parabolic motion is great while the high-altitude parabolic motion is taken as city civilization behavior. Because the implementation places are high floors, few witnesses are available, the parabolic time is short, and even people intentionally hide shadows during parabolic, related departments are difficult to follow the legal responsibility of the parabolic objects, and the events are rare and forbidden frequently. Therefore, the identification and evidence collection of high-altitude parabolic behaviors are an urgent need for city management and property management at present.
The identification and the evidence obtaining of the parabolic behavior can be carried out in a parabolic target track detection mode, the detection of the motion track of the parabolic target is mainly realized in a motion target detection and track fitting mode at present, the method firstly realizes the extraction of the motion target in continuous frames through a classical motion target detection method, such as background modeling or a frame difference method, then fits the track of the motion target in the continuous frames through the coordinates of the motion target in each frame of image, compares the track with the known parabolic track or a parabolic formula to obtain the parabolic track, and finally determines whether the target motion track is the parabolic motion track. In practical application, the method has certain problems, because the roof of a high-rise building is influenced by atmospheric activities, the roof has turbulence, the parabolic targets are various and have different aerodynamic shapes, the motion track of the high-altitude parabolic target is very complex and is difficult to describe by using a theoretical formula, and the comparison with the known high-altitude parabolic track sample is adopted, so that the problem that the track sample is difficult to obtain exists, meanwhile, the whole sample library needs to be traversed during the track comparison, the calculation complexity is high, and the system is complex; and the simulation mode of the experiment is adopted, because the experiment site is limited, the cost of the experiment is high, variable wind field conditions in the actual scene cannot be simulated, the motion tracks of the throwers with different pneumatic shapes under different wind fields are difficult to obtain, and the model lacks effective training sample data in the training process and directly influences the discrimination performance of the model.
Disclosure of Invention
The invention aims to provide a high-altitude tossing object distinguishing method, which converts the high-altitude parabolic target motion detection and identification into a sequence classification problem, improves the detection performance and reduces the complexity of a detection system.
In order to achieve the aim, the invention provides a high-altitude throwing object distinguishing method, which comprises the following steps:
(S1) the camera photographs the floor of the building;
(S2) detecting and extracting a moving object in the camera view using a frame difference method;
(S3) extracting a motion trajectory coordinate sequence of the moving object;
(S4) inputting the motion trail coordinate sequence of the moving target into a pre-trained neural network discrimination model for classification discrimination, and discriminating whether the moving target belongs to a high altitude parabola.
Further, if the moving target belongs to a high-altitude parabola, the motion characteristics of the moving target are added into a training set of a neural network discrimination model, and model iteration is carried out.
Further, the pre-trained neural network discriminant model is obtained by the following steps:
(S41) obtaining motion trail coordinate sequence samples of the throwers with different pneumatic shapes under different simulated wind field conditions through computer simulation;
(S42) taking the throwing object motion trail coordinate sequence sample set as a training set, and training the discriminant neural network to obtain a trained neural network discriminant model.
Further, the method for obtaining the motion trail samples of the throwers with different aerodynamic shapes under different wind field conditions through computer simulation specifically comprises the following steps:
(S411) randomly generating 1 throwing object through computer simulation, wherein the shape of the throwing object is randomly determined in 3 shapes of a block, a rod and a plate;
(S412) initializing the movement state of the throwing object, setting and calculating the iteration time step length delta t and the total iteration step number of the geometric center point coordinate of the throwing object, calculating and recording the geometric center point coordinate on each iteration time step, and forming the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate on each iteration time step into 1 coordinate sequence PtagStoring the data in a database;
(S413) obtaining a motion trail sample P by setting the throwing objects with different pneumatic shapes to move under different wind field conditionstag1、Ptag2···PtagmGenerating 1 trace sample set P ═ Ptag1,Ptag2,...,Ptagm}; optionally, the initializing the movement state of the throwing object specifically includes the following contents: randomly setting initial velocity vector of throwing object
Figure BDA0002763079460000021
The angle of the angle can be selected at will,
Figure BDA0002763079460000022
Figure BDA0002763079460000024
the unit of (a) is m/s; rotational speed of projectile
Figure BDA0002763079460000023
The simulation time is synchronous with the wind field simulation time; the starting point of the movement of the throwing object is the center of the upper edge of the simulated wind field; wherein the windward angle of the plate-shaped throwing object
Figure BDA0002763079460000031
The value is taken randomly,
Figure BDA0002763079460000032
further, the geometric center point coordinate of each iteration time step is calculated and recorded, and the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate of each iteration time step form 1 coordinate sequence PtagStoring the data in a database; the following steps are specifically executed:
(Q1) obtaining at t from the simulated wind farmn-1Wind velocity vector at movement position of throwing object at moment
Figure BDA0002763079460000033
Calculating t by combining the characteristic parameters of the morphology of the throwing objectn-1Surface wind force of throwing object at any moment
Figure BDA0002763079460000034
And
Figure BDA0002763079460000035
wherein the content of the first and second substances,
Figure BDA0002763079460000036
and
Figure BDA0002763079460000037
are each tn-1The surface wind force of the throwing object in the x, y and z directions of 3 dimensions of the space at the moment,
Figure BDA0002763079460000038
is tn-1Turning acting force applied to the throwing object at any moment; wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenRepresenting the current time after the current iteration time step of n times;
(Q2) according to the surface wind force of the throwing
Figure BDA0002763079460000039
And
Figure BDA00027630794600000310
separately calculating the acceleration of the projectile
Figure BDA00027630794600000311
Figure BDA00027630794600000312
And
Figure BDA00027630794600000313
wherein the content of the first and second substances,
Figure BDA00027630794600000314
and
Figure BDA00027630794600000315
respectively the acceleration generated by the force of the throwing object on the space in 3 dimensions x, y and z,
Figure BDA00027630794600000316
acceleration of the projectile in overturning;
(Q3) by tn-1Moment projectile acceleration
Figure BDA00027630794600000317
And tn-1Velocity of the projectile in 3 dimensions x, y and z in space at a time
Figure BDA00027630794600000318
And
Figure BDA00027630794600000319
separately calculate tnVelocity of the projectile in 3 dimensions x, y and z in space at a time
Figure BDA00027630794600000320
And
Figure BDA00027630794600000321
and displacement of
Figure BDA00027630794600000322
And
Figure BDA00027630794600000323
and throw tumbling speed
Figure BDA00027630794600000324
And displacement of
Figure BDA00027630794600000325
And then obtain tnThe coordinates of the geometric center point on the current iteration time step at the moment;
(Q4) when the iteration starts, sequentially iterating according to a set time step delta t, wherein the steps (Q1) to (Q3) are repeatedly executed for each iteration until the total iteration steps are stopped, the geometric center point coordinate of the throwing object on each iteration time step is recorded, and the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate on each iteration time step are usedForm 1 coordinate sequence PtagAnd storing the data in a database.
Further, for the block-shaped throwing objects, spheres, spherical polyhedrons and cubic throwing objects are randomly generated, the diameters of the spheres and the spherical polyhedrons are randomly selected within the range of [10cm,50cm ], and the side lengths of the cubes are randomly selected within the range of [10cm,30cm ];
for the rod-shaped throwing object, randomly generating a cylindrical throwing object and a polygonal throwing object, randomly selecting the length within [30cm,300cm ], and randomly selecting the section diameter within the range of [5cm,20cm ];
for plate shaped throws, rectangular, circular and polygonal plate shaped throws were randomly generated, with areas randomly selected within [0.1m2,1m2 ];
wherein the mass of the throwing substance is randomly selected, m belongs to [50,5000], and the mass unit is g.
Further, the simulated wind field generation steps are as follows:
(D1) setting a calculation space for simulating a wind field, and setting boundary layer conditions of the simulated wind field;
(D2) calculating the boundary layer condition of the set simulation wind field at each calculation point tnTime of day, coordinate xiWind velocity component v (x) of the wind velocity in the x, y, z directioni,tn)、u(xi,tn) And w (x)i,tn) Finally, is obtained at tnThe resultant wind speed V (x) at each calculation point at that momenti,tn)，
Figure BDA0002763079460000041
Wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenRepresenting the current time after the current iteration time step of n times;
(D3) sequentially and iteratively repeating the steps (D1) to (D2) according to a set time step delta t until the total iteration steps are reached, and stopping calculation to obtain a numerical model V of the simulated wind field corresponding to the boundary layer condition of the simulated wind fieldX,T. Alternatively, the x, y,the component v (x) of the wind speed in the z directioni,tn)、u(xi,tn) And w (x)i,tn) By calculating the formula for the wind speed:
Figure BDA0002763079460000042
splitting to obtain v (x) after splittingi,tn)、u(xi,tn) And w (x)i,tn) The formula of (1) is:
Figure BDA0002763079460000043
Figure BDA0002763079460000044
Figure BDA0002763079460000051
for the v direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
For the u direction, R (Δ T) ═ exp (- Δ T/T)Lu')，u”(xi,tn-1)＝σu′[1-R2(Δt)]1/2ξ；
For the w direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
Wherein the content of the first and second substances,
Figure BDA0002763079460000052
and
Figure BDA0002763079460000053
is tnWind speed average component V (x) at timei,tn) The average component of wind speed in the v, u and w directions; v "(x)i,tn-1)、u”(xi,tn-1) And w ″ (x)i,tn-1) Is tn-1The wind velocity pulse component V "(x) at the momenti,tn-1) Wind speed impulse components in the v, u and w directions; xi is 1 group of random numbers which accord with standard normal distribution; sigma is calculated according to the atmospheric state of the boundary layer, sigmav'、σu'And σw'Values of sigma in v, u and w directions respectively; t isLIs tnTime wind speed V (x)i,tn) Pulsating component V' (x)i,tn) Lagrange time scale of (T)Lv'、TLu'And TLw'Are respectively TLLagrange time scales in the v, u, and w directions; Δ t is the calculation time step; r (Δ t) is a correlation coefficient in an exponential form.
Further, the boundary layer conditions include an unstable boundary layer, a neutral boundary layer and a stable boundary layer, and the numerical model V of the simulated wind field corresponding to the unstable boundary layer, the neutral boundary layer and the stable boundary layer is obtained through the steps (D1) to (D3) respectivelyX,T(ii) a Alternatively, for an unstable boundary layer, then σu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝σv′＝u*(12+0.5zi/|L|)1/3；
Figure BDA0002763079460000054
TLu′＝TLv′＝0.15zi/σu′；
Figure BDA0002763079460000061
for a neutral boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*exp(-3fz/u*)；
σv′＝σw′＝1.3u*exp(-2fz/u*)；
Figure BDA0002763079460000062
for a stable boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*(1-z/zi)；
σv′＝σw′＝1.3u*(1-z/zi)；
Figure BDA0002763079460000063
Figure BDA0002763079460000064
Figure BDA0002763079460000065
wherein z is the calculated point height, u*As the friction speed, w*Is the convection characteristic velocity. z is a radical ofiThe height of the mixed layer is shown, L is the length of the Morin-obufhoff, and the value is taken according to the boundary condition. f is the Coriolis force length, and 7.29 multiplied by 10 are taken according to the latitude distribution of China-5。
Further, the training set construction step comprises:
(S421) simulating a camera position in a simulated environment according to the position of the camera layout in the actual use environment, wherein the included angle between the optical axis of the simulated camera and the simulated floor is equal to the included angle between the optical axis of the camera and the floor in the actual environment;
(S422) simulating the motion track of the throwing object to obtain track images of various throwing objects in the simulation camera under various throwing object forms and various wind field conditions;
(S423) simulating the motion trail of the non-throwing object, randomly generating a transverse flying target, a longitudinal ascending target, a transverse reciprocating target, a longitudinal reciprocating target and a random flashing target, recording the motion trail of the transverse flying target, the longitudinal ascending target, the transverse reciprocating target, the longitudinal reciprocating target and the random flashing target, and obtaining the motion trail images of various non-throwing object motion targets under various types of non-throwing objects and various wind field conditions;
(S424) mixing the throwing object motion track image sample with the non-throwing object motion track image sample according to the proportion of 1: 1;
(S425) mixing the mixed sample set by the number of training samples: and (5) sampling according to the ratio of 7:3 of the number of the test samples, and constructing a training data set and a test data set.
Further, the throwing object and the non-throwing object can obtain the motion tracks of various types of throwing objects or non-throwing objects and various wind field conditions through the following steps:
(F1) let the coordinates of the camera be (x)c,yc,zc) The motion track of the non-throwing object or throwing object is obtained through simulation, and the coordinate of the non-throwing object or throwing object at the time t is set as (x)tag,ytag,ztag) Coordinates (x ') of non-projectile or projectile track points in the camera image plane'tag,y'tag，z'tag)，x'tagAnd y'tagThe calculation formulas of (A) and (B) are respectively as follows:
Figure BDA0002763079460000071
Figure BDA0002763079460000072
Figure BDA0002763079460000073
wherein f is the focal length of the camera lens;
(F2) repeating the step (F1) until all non-throws or throws position points on 1 track are imaged in the analog camera and form a track image, and then performing the step (F3);
(F3) repeating the step (F1) and the step (F2) to obtain track images of various non-throws or throws in the analog camera under various wind field conditions of various throws; optionally, the neural network discrimination model is obtained by convolutional neural network training; optionally, the neural network discrimination model is divided into an input layer, a feature extraction layer and a discrimination output layer; the input layer receives an input motion track coordinate sequence and feeds the motion track coordinate sequence to the feature extraction layer; the characteristic extraction layer is formed by stacking 3 convolution pooling layers, namely, 1 dimension pooling layer is connected behind each convolution layer; the judgment output layer is composed of 1 full-connection layer and 1 softmax layer, and softmax carries out two-classification judgment, namely whether the input track is the high-altitude throwing object motion track or not; optionally, the length of the input motion trajectory coordinate sequence is 125, the sequences exceeding 125 retain data from the 1 st bit to 125 th bit, and the sequences less than 125 complement 0 at the end of the sequence; optionally, in each convolution pooling layer, the size of a convolution kernel of the convolution layer is 1 × 5, and the step length of the convolution kernel is 1; the pooling size was 1 × 5 with a step size of 2.
Compared with the prior art, the invention has the following advantages:
the high-altitude throwing object distinguishing system converts the high-altitude throwing object motion detection and identification into a sequence classification problem, improves the detection performance and reduces the complexity of the detection system; the problem that a track sample is difficult to obtain is solved; the system is simple, calculation is greatly reduced, cost is reduced, changeable wind field conditions in an actual scene can be simulated, the motion tracks of throwers with different pneumatic shapes under different wind fields can be obtained, effective training sample data can be obtained by the model in the training process, and the distinguishing performance of the model is improved.
Drawings
FIG. 1 is a flow chart of a method for discriminating a high altitude tossing object according to the present invention;
FIG. 2 is a schematic structural diagram for high-altitude tossing discrimination according to the present invention.
In the figure:
1-video image acquisition module; 2-moving object detection module; 3-an image processing module; 41-model iteration module; 4-high altitude parabolic discrimination module; 5-network transmission module.
Detailed Description
The following further describes embodiments of the present invention with reference to the drawings.
Referring to fig. 1, the present embodiment discloses a method for discriminating a high-altitude tossing object, which includes the following steps:
(S1) the camera photographs the floor of the building;
(S2) detecting and extracting a moving object in the camera view using a frame difference method;
(S3) extracting a motion trajectory coordinate sequence of the moving object;
(S4) inputting the motion trail coordinate sequence of the moving target into a pre-trained neural network discrimination model, and discriminating whether the moving target belongs to a high altitude parabola.
In this embodiment, if the moving object belongs to a high-altitude parabola, the motion characteristics of the moving object are added to a training set of a neural network discriminant model, and model iteration is performed.
In this embodiment, the pre-trained neural network discriminant model is obtained by:
(S41) obtaining motion trail coordinate sequence samples of the throwers with different pneumatic shapes under different simulated wind field conditions through computer simulation;
(S42) taking the throwing object motion trail coordinate sequence sample set as a training set, and training the discriminant neural network to obtain a trained neural network discriminant model.
In this embodiment, the obtaining of the motion trajectory samples of the throwers with different aerodynamic shapes under different wind field conditions through computer simulation specifically includes the following steps:
(S411) randomly generating 1 throwing object through computer simulation, wherein the shape of the throwing object is randomly determined in 3 shapes of a block, a rod and a plate;
(S412) initializing the motion state of the throwing object and setting and calculating the coordinates of the geometric center point of the throwing objectIteration time step length delta t and total iteration step number, calculating and recording the coordinate of the geometric center point on each iteration time step, and forming the coordinate of the geometric center point of the throwing object in an initial state and the coordinate of the geometric center point on each iteration time step into 1 coordinate sequence PtagStoring the data in a database;
(S413) obtaining a motion trail sample P by setting the throwing objects with different pneumatic shapes to move under different wind field conditionstag1、Ptag2...PtagmGenerating 1 trace sample set P ═ Ptag1,Ptag2,...,Ptagm}。
In this embodiment, the initializing the throwing object motion state specifically includes the following: randomly setting initial velocity vector of throwing object
Figure BDA0002763079460000091
The angle of the angle can be selected at will,
Figure BDA0002763079460000092
Figure BDA0002763079460000093
the unit of (a) is m/s; rotational speed of projectile
Figure BDA0002763079460000094
The simulation time is synchronous with the wind field simulation time;
the starting point of the movement of the throwing object is the center of the upper edge of the simulated wind field. Wherein the windward angle of the plate-shaped throwing object
Figure BDA0002763079460000095
The value is taken randomly,
Figure BDA0002763079460000096
windward angle of plate-shaped throwing object
Figure BDA0002763079460000097
Is used for inputting Fluent calculation.
In this embodiment, the sum is calculatedRecording the coordinate of the geometric center point on each iteration time step, and forming the coordinate of the geometric center point of the throwing object in the initial state and the coordinate of the geometric center point on each iteration time step into 1 coordinate sequence PtagStoring the data in a database; the following steps are specifically executed:
(Q1) obtaining at t from the simulated wind farmn-1Wind velocity vector at movement position of throwing object at moment
Figure BDA0002763079460000101
Calculating t by combining the characteristic parameters of the morphology of the throwing objectn-1Surface wind force of throwing object at any moment
Figure BDA0002763079460000102
And
Figure BDA0002763079460000103
wherein the content of the first and second substances,
Figure BDA0002763079460000104
and
Figure BDA0002763079460000105
are each tn-1The surface wind force of the throwing object in the x, y and z directions of 3 dimensions of the space at the moment,
Figure BDA0002763079460000106
is tn-1Turning acting force applied to the throwing object at any moment; wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenRepresenting the current time after the current iteration time step of n times;
(Q2) according to the surface wind force of the throwing
Figure BDA0002763079460000107
And
Figure BDA0002763079460000108
separately calculating the acceleration of the projectile
Figure BDA0002763079460000109
Figure BDA00027630794600001010
And
Figure BDA00027630794600001011
wherein the content of the first and second substances,
Figure BDA00027630794600001012
and
Figure BDA00027630794600001013
respectively the acceleration generated by the force of the throwing object on the space in 3 dimensions x, y and z,
Figure BDA00027630794600001014
acceleration of the projectile in overturning;
(Q3) by tn-1Moment projectile acceleration
Figure BDA00027630794600001015
And tn-1Velocity of the projectile in 3 dimensions x, y and z in space at a time
Figure BDA00027630794600001016
And
Figure BDA00027630794600001017
separately calculate tnVelocity of the projectile in 3 dimensions x, y and z in space at a time
Figure BDA00027630794600001018
And
Figure BDA00027630794600001019
and displacement of
Figure BDA00027630794600001020
And
Figure BDA00027630794600001021
and throw tumbling speed
Figure BDA00027630794600001022
And displacement of
Figure BDA00027630794600001023
And then obtain tnThe coordinates of the geometric center point on the current iteration time step at the moment;
(Q4) when iteration starts, sequentially iterating according to a set time step delta t, wherein the steps (Q1) to (Q3) are repeatedly executed for each iteration until the total iteration steps are stopped, the geometric center point coordinate of the throwing object on each iteration time step is recorded, and the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate on each iteration time step form 1 coordinate sequence PtagStoring the data in a database; wherein, when the preset iteration step number is reached, the throwing object falls to the ground or flies out of the boundary of the simulated wind field; the throwing object falling to the ground is Vtag,y,t＝0。
In this embodiment, for the block-shaped thrower, a sphere, a spherical polyhedron and a cubic thrower are randomly generated, the diameter of the sphere and the spherical polyhedron is randomly selected within the range of [10cm,50cm ], and the side length of the cube is randomly selected within the range of [10cm,30cm ];
for the rod-shaped throwing object, randomly generating a cylindrical throwing object and a polygonal throwing object, randomly selecting the length within [30cm,300cm ], and randomly selecting the section diameter within the range of [5cm,20cm ];
for plate shaped throws, rectangular, circular and polygonal plate shaped throws were randomly generated, with areas randomly selected within [0.1m2,1m2 ];
wherein the mass of the throwing substance is randomly selected, m belongs to [50,5000], and the mass unit is g.
In the present embodiment, the surface wind force
Figure BDA0002763079460000111
And
Figure BDA0002763079460000112
the calculation formulas of (A) and (B) are respectively as follows:
Figure BDA0002763079460000113
Figure BDA0002763079460000114
Figure BDA0002763079460000115
Figure BDA0002763079460000116
where ρ isaIs the air density, A is the maximum reference area of the projectile,
Figure BDA0002763079460000117
is tn-1The velocity vector of the projectile motion at the moment,
Figure BDA0002763079460000118
is the projectile rotation velocity vector;
Figure BDA0002763079460000119
is tn-1A wind speed vector at the movement position of the throwing object at the moment;
Figure BDA00027630794600001110
and
Figure BDA00027630794600001111
are each tn-1The surface wind power coefficient of the throwing object in the x, y and z directions of 3 dimensions of the space at the moment,
Figure BDA00027630794600001112
is tn-1The throwing object is turned over at any momentThe force corresponds to the surface wind coefficient. In this embodiment, at each iteration, the following data needs to be updated: the position coordinates of the projectile; velocity vector of projectile
Figure BDA00027630794600001113
Wind velocity vector at current motion position of throwing object
Figure BDA00027630794600001114
Throwing at tn-1Surface wind coefficient of throwing object at any moment
Figure BDA00027630794600001115
And
Figure BDA00027630794600001116
Figure BDA00027630794600001117
and
Figure BDA00027630794600001118
is according to tn-1Wind velocity vector at movement position of throwing object at moment
Figure BDA00027630794600001119
And calculating by using an RANS (random average time stress) model and CFD (computational fluid dynamics) calculation software Fluent according to the projectile attitude characteristic parameters. t is tnSurface wind force of throwing object at any moment
Figure BDA00027630794600001120
And
Figure BDA00027630794600001121
and calculating by using an RANS (random average time stress) model and CFD (computational fluid dynamics) calculation software Fluent.
In this embodiment, the projectile acceleration is calculated
Figure BDA00027630794600001122
And
Figure BDA00027630794600001123
respectively as follows:
Figure BDA00027630794600001124
Figure BDA0002763079460000121
Figure BDA0002763079460000122
Figure BDA0002763079460000123
calculating the velocity of the projectile in 3 dimensions x, y and z in space
Figure BDA0002763079460000124
And
Figure BDA0002763079460000125
and displacement of
Figure BDA0002763079460000126
Figure BDA0002763079460000127
And
Figure BDA0002763079460000128
and throw tumbling speed
Figure BDA0002763079460000129
And displacement of
Figure BDA00027630794600001210
Respectively as follows:
Figure BDA00027630794600001211
Figure BDA00027630794600001212
Figure BDA00027630794600001213
Figure BDA00027630794600001214
Figure BDA00027630794600001215
Figure BDA00027630794600001216
Figure BDA00027630794600001217
Figure BDA00027630794600001218
in this embodiment, the simulated wind field generating step is as follows:
(D1) setting a calculation space for simulating a wind field, and setting boundary layer conditions of the simulated wind field;
(D2) calculating the boundary layer condition of the set simulation wind field at each calculation point tnTime of day, coordinate xiWind velocity component v (x) of the wind velocity in the x, y, z directioni,tn)、u(xi,tn) And w (x)i,tn) Finally, is obtained at tnThe resultant wind speed V (x) at each calculation point at that momenti,tn)，
Figure BDA00027630794600001219
Wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenRepresenting the current time after the current iteration time step of n times;
(D3) sequentially and iteratively repeating the steps (D1) to (D2) according to a set time step delta t until the total iteration steps are reached, and stopping calculation to obtain a numerical model V of the simulated wind field corresponding to the boundary layer condition of the simulated wind fieldX,T。
In this embodiment, the size of the computation space is 120m in height, 300m in width, and 300m in depth; in some embodiments, the computation space may be other sizes, and is not limited herein. Dividing each calculation point: according to the computing power, the grid size of the wind field computing points is determined, and the simulation grid size is selectable within 1mX1m to 5mX5 m. The wind field simulation of the unstable boundary layer selects a small-scale and dense grid, and the simulation of the neutral and stable boundary layers selects a large-scale and sparse grid. And confirming that the simulation time of the wind field is the same as the total iteration steps under the movement state of the throwing object.
In the present embodiment, it is preferred that,
Figure BDA0002763079460000131
at [0m/s,5m/s]And randomly setting.
In the present embodiment, the wind speed component v (x) in the x, y, z direction can be derived by the formula (1) to the formula (4)i,tn)、u(xi,tn) And w (x)i,tn) The specific calculation process of (2) is as follows:
simulating the wind speed of each calculation point in the wind field at the time t:
Figure BDA0002763079460000132
equation (1) can be broken down into 3 velocity components to represent:
Figure BDA0002763079460000133
the formula for generating the synthetic wind speed by the 3 velocity components is as follows:
Figure BDA0002763079460000134
taking the airflow vortex process as 1 continuous process, and calculating the point wind speed by following 1 time step under the Markov assumption condition as follows:
V′(xi,tn)＝R(Δt)V′(xi,tn-1)+V″(xi,tn-1) (2)；
equation (2) can be broken down into 3 velocity components to represent:
Figure BDA0002763079460000141
R(Δt)＝exp(-Δt/TL) (3)；
V″(xi,tn)＝σ[1-R2(Δt)]1/2ξ (4)；
to sum up, a wind speed calculation formula is obtained:
Figure BDA0002763079460000142
wind velocity component v (x) in the x, y, z directioni,tn)、u(xi,tn) And w (x)i,tn) V (x) after resolutioni,tn)、u(xi,tn) And w (x)i,tn) The formula of (1) is:
Figure BDA0002763079460000143
Figure BDA0002763079460000144
Figure BDA0002763079460000145
for the v direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
For the u direction, R (Δ T) ═ exp (- Δ T/T)Lu')，u”(xi,tn-1)＝σu′[1-R2(Δt)]1/2ξ；
For the w direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
Wherein the content of the first and second substances,
Figure BDA0002763079460000146
and
Figure BDA0002763079460000147
is tnWind speed average component V (x) at timei,tn) The average component of wind speed in the v, u and w directions; v "(x)i,tn-1)、u”(xi,tn-1) And w ″ (x)i,tn-1) Is tn-1The wind velocity pulse component V "(x) at the momenti,tn-1) Wind speed impulse components in the v, u and w directions; xi is 1 group of random numbers which accord with standard normal distribution; sigma is calculated according to the atmospheric state of the boundary layer, sigmav'、σu'And σw'Values of sigma in v, u and w directions respectively; t isLIs tnTime wind speed V (x)i,tn) Pulsating component V' (x)i,tn) Lagrange time scale of (T)Lv'、TLu'And TLw'Are respectively TLLagrange time scales in the v, u, and w directions; Δ t is the calculation time step; r (Δ t) is a correlation coefficient in an exponential form. The mean component is in principle given here by 1 piece of diagnostic wind field softwareHowever, since the parabolic time is short and the average component of the wind speed does not change much, in the present scheme, the average component of the wind speed is artificially set and kept constant in each simulation calculation, and therefore
Figure BDA0002763079460000148
And
Figure BDA0002763079460000149
is taken from the value of [1,10]In the interval of the time interval,
Figure BDA00027630794600001410
is taken to be [0,3 ]]Within the interval. The time step is taken according to actual needs, and is not limited herein, and in this embodiment, the time step is taken as 1 s. Computationally, the second term on the middle right of the above 3 equations requires the pair v' (x)i,tn)、u'(xi,tn)、w'(xi,tn) And (6) performing iteration. When the above three equations are used for description. In actual calculations V' has an initial value of 0 or 1 very small value.
In the present embodiment, the boundary layer conditions include an unstable boundary layer, a neutral boundary layer and a stable boundary layer, and the numerical models V of the simulated wind fields corresponding to the unstable boundary layer, the neutral boundary layer and the stable boundary layer are obtained through the steps (D1) to (D3), respectivelyX,T。
In this embodiment, for an unstable boundary layer, then σu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝σv′＝u*(12+0.5zi/|L|)1/3；
Figure BDA0002763079460000151
TLu′＝TLv′＝0.15zi/σu′；
Figure BDA0002763079460000152
for a neutral boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*exp(-3fz/u*)；
σv′＝σw′＝1.3u*exp(-2fz/u*)；
Figure BDA0002763079460000153
for a stable boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*(1-z/zi)；
σv′＝σw′＝1.3u*(1-z/zi)；
Figure BDA0002763079460000161
Figure BDA0002763079460000162
Figure BDA0002763079460000163
wherein z is the calculated point height, u*As the friction speed, w*Is the convection characteristic velocity. z is a radical ofiThe height of the mixed layer is L is the length of Monin-Obukhov, and the value is taken according to the boundary condition. f is the Coriolis force length, and 7.29 multiplied by 10 are taken according to the latitude distribution of China-5。
In this embodiment, the training set constructing step includes:
(S421) simulating a camera position in a simulated environment according to the position of the camera layout in the actual use environment, wherein the included angle between the optical axis of the simulated camera and the simulated floor is equal to the included angle between the optical axis of the camera and the floor in the actual environment;
(S422) simulating the motion track of the throwing object to obtain track images of various throwing objects in the simulation camera under various throwing object forms and various wind field conditions;
(S423) simulating the motion trail of the non-throwing object, randomly generating a transverse flying target, a longitudinal ascending target, a transverse reciprocating target, a longitudinal reciprocating target and a random flashing target, recording the motion trail of the transverse flying target, the longitudinal ascending target, the transverse reciprocating target, the longitudinal reciprocating target and the random flashing target, and obtaining the motion trail images of various non-throwing object motion targets under various types of non-throwing objects and various wind field conditions;
(S424) mixing the throwing object motion track image sample with the non-throwing object motion track image sample according to the proportion of 1: 1;
(S425) mixing the mixed sample set by the number of training samples: and (5) sampling according to the ratio of 7:3 of the number of the test samples, and constructing a training data set and a test data set.
In the embodiment, the throwing object and the non-throwing object are all subjected to the following steps to obtain the motion tracks of various types of throwing objects or non-throwing objects and various wind field conditions:
(F1) let the coordinates of the camera be (x)c,yc,zc) The motion track of the non-throwing object or throwing object is obtained through simulation, and the coordinate of the non-throwing object or throwing object at the time t is set as (x)tag,ytag,ztag) Coordinates (x ') of non-projectile or projectile track points in the camera image plane'tag,y'tag，z'tag)，x'tagAnd y'tagThe calculation formulas of (A) and (B) are respectively as follows:
Figure BDA0002763079460000171
Figure BDA0002763079460000172
Figure BDA0002763079460000173
wherein f is the focal length of the camera lens; the pinhole imaging model is used, in which no calculation is performed on the data in the Z-axis.
(F2) Repeating the step (F1) until all non-throws or throws position points on 1 track are imaged in the analog camera and form a track image, and then performing the step (F3);
(F3) and (F1) repeating the steps (F2) to obtain track images of different non-throws or throws in the analog camera under various throwing forms and wind field conditions.
In this embodiment, the neural network discrimination model is obtained by convolutional neural network training. The neural network discrimination model is divided into an input layer, a feature extraction layer and a discrimination output layer; the input layer receives an input coordinate sequence track coordinate sequence and feeds the input coordinate sequence track coordinate sequence to the feature extraction layer; the characteristic extraction layer is formed by stacking 3 convolution pooling layers, namely, 1 dimension pooling layer is connected behind each convolution layer; and the judgment output layer consists of 1 full-connection layer and 1 softmax layer, and the softmax carries out two-classification judgment, namely whether the input track is the high-altitude throwing motion track or not.
In the embodiment, the length of the input motion trail coordinate sequence is 125, sequences exceeding 125 retain data from 1 st bit to 125 th bit, and sequences less than 125 complement 0 at the end of the sequence; optionally, in each convolution pooling layer, the size of a convolution kernel of the convolution layer is 1 × 5, and the step length of the convolution kernel is 1; the pooling size was 1 × 5 with a step size of 2.
In this example, the neural network uses Binary Cross entry as a loss function:
Figure BDA0002763079460000174
in the formula:
y is the category of the track sample;
p (yi) sample class classification probabilities given for the neural network;
and N is the number of samples. The motion track of the throwing object in the wind field is discretized into 1 track coordinate sequence after being shot by a camera. The sequence is input into 1 discrimination network to realize the classification discrimination of the track.
Referring to fig. 2, the embodiment further discloses a high-altitude tossing object determining system, which executes the high-altitude tossing object determining method, and the method includes:
the video image acquisition module 1 is used for shooting the building floor;
the network transmission module 5 is used for transmitting pictures for shooting the floors of the building;
the moving object detection module 2 is used for receiving the pictures of the building floors from the network transmission module 5 and detecting and extracting moving objects in the shot pictures; the moving target detection module 2 is connected with the video image acquisition module 1;
the image processing module 3 is used for extracting a motion track coordinate sequence of the moving target and inputting the motion track coordinate sequence into the high-altitude parabolic judging module 4; the image processing module 3 is connected with the moving object detection module 2;
the high-altitude parabolic judging module 4 is used for judging whether the input moving target belongs to a high-altitude parabolic object, the high-altitude parabolic judging module 4 is connected with the image processing module 3, and a neural network judging model for judging whether the input moving target belongs to the high-altitude parabolic object is established in the high-altitude parabolic judging module 4.
In this embodiment, the high altitude parabolic determination module 4 includes a model iteration module 41, configured to add the motion characteristics of the moving object belonging to the high altitude parabolic model into a training set of the neural network determination model, and perform model iteration.
The high-altitude throwing object distinguishing system converts the high-altitude throwing object motion detection and identification into a sequence classification problem, improves the detection performance and reduces the complexity of the detection system; the problem that a track sample is difficult to obtain is solved; the system is simple, calculation is greatly reduced, cost is reduced, changeable wind field conditions in an actual scene can be simulated, the motion tracks of throwers with different pneumatic shapes under different wind fields can be obtained, effective training sample data can be obtained by the model in the training process, and the distinguishing performance of the model is improved.
The foregoing detailed description of the preferred embodiments of the invention has been presented. It should be understood that numerous modifications and variations could be devised by those skilled in the art in light of the present teachings without departing from the inventive concepts. Therefore, the technical solutions available to those skilled in the art through logic analysis, reasoning and limited experiments based on the prior art according to the concept of the present invention should be within the scope of protection defined by the claims.

Claims (10)

1. A high-altitude throwing object distinguishing method is characterized by comprising the following steps:
(S1) the camera photographs the floor of the building;
(S2) detecting and extracting a moving object in the camera view using a frame difference method;
(S3) extracting a motion trajectory coordinate sequence of the moving object;
(S4) inputting the motion trail coordinate sequence of the moving target into a pre-trained neural network discrimination model for classification discrimination, and discriminating whether the moving target belongs to a high altitude parabola.
2. The high-altitude tossing object discrimination method according to claim 1, wherein the pre-trained neural network discrimination model is obtained by:
(S41) obtaining motion trail coordinate sequence samples of the throwers with different pneumatic shapes under different simulated wind field conditions through computer simulation;
(S42) taking the throwing object motion trail coordinate sequence sample set as a training set, and training the discriminant neural network to obtain a trained neural network discriminant model.
3. The high-altitude tossing object distinguishing method according to claim 2, wherein the motion track samples of the tossing objects with different aerodynamic shapes under different wind field conditions are obtained through computer simulation, and the method specifically comprises the following steps:
(S411) randomly generating 1 throwing object through computer simulation, wherein the shape of the throwing object is randomly determined in 3 shapes of a block, a rod and a plate;
(S412) initializing the movement state of the throwing object, setting and calculating the iteration time step length delta t and the total iteration step number of the geometric center point coordinate of the throwing object, calculating and recording the geometric center point coordinate on each iteration time step, and forming the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate on each iteration time step into 1 coordinate sequence PtagStoring the data in a database;
(S413) obtaining a motion trail sample P by setting the throwing objects with different pneumatic shapes to move under different wind field conditionstag1、Ptag2···PtagmGenerating 1 trace sample set P ═ Ptag1,Ptag2,...,Ptagm}; optionally, the initializing the movement state of the throwing object specifically includes the following contents: randomly setting initial velocity vector of throwing object
Figure FDA0002763079450000011
The angle of the angle can be selected at will,
Figure FDA0002763079450000012
Figure FDA0002763079450000013
the unit of (a) is m/s; rotational speed of projectile
Figure FDA0002763079450000014
The simulation time is synchronous with the wind field simulation time; the starting point of the movement of the throwing object is the center of the upper edge of the simulated wind field; wherein the windward angle of the plate-shaped throwing object
Figure FDA0002763079450000021
The value is taken randomly,
Figure FDA0002763079450000022
4. the method for discriminating a high altitude tossing object according to claim 3, wherein the geometric center point coordinates at each iteration time step are calculated and recorded, and the geometric center point coordinates of the tossing object at the initial state and the geometric center point coordinates at each iteration time step are combined into 1 coordinate sequence PtagStoring the data in a database; the following steps are specifically executed:
(Q1) obtaining at t from the simulated wind farmn-1Wind velocity vector at movement position of throwing object at moment
Figure FDA0002763079450000023
Calculating t by combining the characteristic parameters of the morphology of the throwing objectn-1Surface wind force of throwing object at any moment
Figure FDA0002763079450000024
And
Figure FDA0002763079450000025
wherein the content of the first and second substances,
Figure FDA0002763079450000026
and
Figure FDA0002763079450000027
are each tn-1The surface wind force of the throwing object in the x, y and z directions of 3 dimensions of the space at the moment,
Figure FDA0002763079450000028
is tn-1Turning acting force applied to the throwing object at any moment; wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenIndicates the currentThe current time after n iteration time steps;
(Q2) according to the surface wind force of the throwing
Figure FDA0002763079450000029
And
Figure FDA00027630794500000210
separately calculating the acceleration of the projectile
Figure FDA00027630794500000211
Figure FDA00027630794500000212
And
Figure FDA00027630794500000213
wherein the content of the first and second substances,
Figure FDA00027630794500000214
and
Figure FDA00027630794500000215
respectively the acceleration generated by the force of the throwing object on the space in 3 dimensions x, y and z,
Figure FDA00027630794500000216
acceleration of the projectile in overturning;
(Q3) by tn-1Moment projectile acceleration
Figure FDA00027630794500000217
And tn-1Velocity of the projectile in 3 dimensions x, y and z in space at a time
Figure FDA00027630794500000218
And
Figure FDA00027630794500000219
separately calculate tnVelocity of the projectile in 3 dimensions x, y and z in space at a time
Figure FDA00027630794500000220
And
Figure FDA00027630794500000221
and displacement of
Figure FDA00027630794500000222
And
Figure FDA00027630794500000223
and throw tumbling speed
Figure FDA00027630794500000224
And displacement of
Figure FDA00027630794500000225
And then obtain tnThe coordinates of the geometric center point on the current iteration time step at the moment;
(Q4) when iteration starts, sequentially iterating according to a set time step delta t, wherein the steps (Q1) to (Q3) are repeatedly executed for each iteration until the total iteration steps are stopped, the geometric center point coordinate of the throwing object on each iteration time step is recorded, and the geometric center point coordinate of the throwing object in the initial state and the geometric center point coordinate on each iteration time step form 1 coordinate sequence PtagAnd storing the data in a database.
5. The high altitude tossing object discrimination method according to claim 3 or 4, wherein,
for the blocky throwing object, randomly generating a sphere, a spherical polyhedron and a cubic throwing object, wherein the diameter of the sphere and the spherical polyhedron is randomly selected within the range of [10cm,50cm ], and the side length of the cube is randomly selected within the range of [10cm,30cm ];
for the rod-shaped throwing object, randomly generating a cylindrical throwing object and a polygonal throwing object, randomly selecting the length within [30cm,300cm ], and randomly selecting the section diameter within the range of [5cm,20cm ];
for plate shaped throws, rectangular, circular and polygonal plate shaped throws were randomly generated, with areas randomly selected within [0.1m2,1m2 ];
wherein the mass of the throwing substance is randomly selected, m belongs to [50,5000], and the mass unit is g.
6. The method for discriminating a high altitude tossing object according to any one of claims 2 to 4, wherein the step of generating the simulated wind field is as follows:
(D1) setting a calculation space for simulating a wind field, and setting boundary layer conditions of the simulated wind field;
(D2) calculating the boundary layer condition of the set simulation wind field at each calculation point tnTime of day, coordinate xiWind velocity component v (x) of the wind velocity in the x, y, z directioni,tn)、u(xi,tn) And w (x)i,tn) Finally, is obtained at tnThe resultant wind speed V (x) at each calculation point at that momenti,tn)，
Figure FDA0002763079450000031
Wherein, tn＝t0+ n Δ t, n is greater than or equal to 1 and less than or equal to the total number of iteration steps; when n is 1, tn-1＝t0，t0Denotes the initial time, t, of the initial state of the projectilenRepresenting the current time after the current iteration time step of n times;
(D3) sequentially and iteratively repeating the steps (D1) to (D2) according to a set time step delta t until the total iteration steps are reached, and stopping calculation to obtain a numerical model V of the simulated wind field corresponding to the boundary layer condition of the simulated wind fieldX,T(ii) a Optionally, the wind velocity component v (x) in the x, y, z directioni,tn)、u(xi,tn) And w (x)i,tn) By calculating the formula for the wind speed:
Figure FDA0002763079450000032
splitting to obtain v (x) after splittingi,tn)、u(xi,tn) And w (x)i,tn) Respectively as follows:
Figure FDA0002763079450000033
Figure FDA0002763079450000041
Figure FDA0002763079450000042
for the v direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
For the u direction, R (Δ T) ═ exp (- Δ T/T)Lu')，u”(xi,tn-1)＝σu′[1-R2(Δt)]1/2ξ；
For the w direction, R (Δ T) ═ exp (- Δ T/T)Lv')，v”(xi,tn-1)＝σv′[1-R2(Δt)]1/2ξ；
Wherein the content of the first and second substances,
Figure FDA0002763079450000043
and
Figure FDA0002763079450000044
is tnWind speed average component V (x) at timei,tn) The average component of wind speed in the v, u and w directions; v "(x)i,tn-1)、u”(xi,tn-1) And w ″ (x)i,tn-1) Is tn-1The wind velocity pulse component V "(x) at the momenti,tn-1) Wind speed impulse components in the v, u and w directions; xi is 1 group of random numbers which accord with standard normal distribution; σ is according to boundary layer atmosphereThe state is calculated to obtainv'、σu'And σw'Values of sigma in v, u and w directions respectively; t isLIs tnTime wind speed V (x)i,tn) Pulsating component V' (x)i,tn) Lagrange time scale of (T)Lv'、TLu'And TLw'Are respectively TLLagrange time scales in the v, u, and w directions; Δ t is the calculation time step; r (Δ t) is a correlation coefficient in an exponential form.
7. The high altitude tossing method as claimed in claim 6, wherein the boundary layer conditions include an unstable boundary layer, a neutral boundary layer and a stable boundary layer, and numerical models V of simulated wind fields corresponding to the unstable boundary layer, the neutral boundary layer and the stable boundary layer are obtained through the steps (D1) to (D3) respectivelyX,T(ii) a Alternatively, for an unstable boundary layer, then σu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝σv′＝u*(12+0.5zi/|L|)1/3；
Figure FDA0002763079450000045
TLu′＝TLv′＝0.15zi/σu′；
Figure FDA0002763079450000051
for a neutral boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*exp(-3fz/u*)；
σv′＝σw′＝1.3u*exp(-2fz/u*)；
Figure FDA0002763079450000052
for a stable boundary layer, thenu'、σv'、σw'、TLu'、TLv'And TLw'Respectively as follows:
σu′＝2u*(1-z/zi)；
σv′＝σw′＝1.3u*(1-z/zi)；
Figure FDA0002763079450000053
Figure FDA0002763079450000054
Figure FDA0002763079450000055
wherein z is the calculated point height, u*As the friction speed, w*Is the convection characteristic velocity. z is a radical ofiThe height of the mixed layer is shown, L is the length of the Morin-obufhoff, and the value is taken according to the boundary condition. f is the Coriolis force length, and 7.29 multiplied by 10 are taken according to the latitude distribution of China-5。
8. The method of discriminating a high altitude tossing object according to claim 2, 3, 4 or 7 wherein the step of constructing a training set comprises:
(S421) simulating a camera position in a simulated environment according to the position of the camera layout in the actual use environment, wherein the included angle between the optical axis of the simulated camera and the simulated floor is equal to the included angle between the optical axis of the camera and the floor in the actual environment;
(S422) simulating the motion track of the throwing object to obtain track images of various throwing objects in the simulation camera under various throwing object forms and various wind field conditions;
(S423) simulating the motion trail of the non-throwing object, randomly generating a transverse flying target, a longitudinal ascending target, a transverse reciprocating target, a longitudinal reciprocating target and a random flashing target, recording the motion trail of the transverse flying target, the longitudinal ascending target, the transverse reciprocating target, the longitudinal reciprocating target and the random flashing target, and obtaining the motion trail images of various non-throwing object motion targets under various types of non-throwing objects and various wind field conditions;
(S424) mixing the throwing object motion track image sample with the non-throwing object motion track image sample according to the proportion of 1: 1;
(S425) mixing the mixed sample set by the number of training samples: and (5) sampling according to the ratio of 7:3 of the number of the test samples, and constructing a training data set and a test data set.
9. The method for discriminating high altitude tossing objects according to claim 8, wherein the tossing objects and the non-tossing objects are each obtained by the following steps:
(F1) let the coordinates of the camera be (x)c,yc,zc) The motion track of the non-throwing object or throwing object is obtained through simulation, and the coordinate of the non-throwing object or throwing object at the time t is set as (x)tag,ytag,ztag) Coordinates (x ') of non-projectile or projectile track points in the camera image plane'tag,y′tag，z′tag)，x′tagAnd y'tagThe calculation formulas of (A) and (B) are respectively as follows:
Figure FDA0002763079450000061
Figure FDA0002763079450000062
Figure FDA0002763079450000063
wherein f is the focal length of the camera lens;
(F2) repeating the step (F1) until all non-throws or throws position points on 1 track are imaged in the analog camera and form a track image, and then performing the step (F3);
(F3) repeating the step (F1) and the step (F2) to obtain track images of various non-throws or throws in the analog camera under various wind field conditions of various throws; optionally, the neural network discrimination model is obtained by convolutional neural network training; optionally, the neural network discrimination model is divided into an input layer, a feature extraction layer and a discrimination output layer; the input layer receives an input motion track coordinate sequence and feeds the motion track coordinate sequence to the feature extraction layer; the characteristic extraction layer is formed by stacking 3 convolution pooling layers, namely, 1 dimension pooling layer is connected behind each convolution layer; the judgment output layer is composed of 1 full-connection layer and 1 softmax layer, and softmax carries out two-classification judgment, namely whether the input track is the high-altitude throwing object motion track or not; optionally, the length of the input motion trajectory coordinate sequence is 125, the sequences exceeding 125 retain data from the 1 st bit to 125 th bit, and the sequences less than 125 complement 0 at the end of the sequence; optionally, in each convolution pooling layer, the size of a convolution kernel of the convolution layer is 1 × 5, and the step length of the convolution kernel is 1; the pooling size was 1 × 5 with a step size of 2.
10. The method for discriminating a high altitude toss according to claim 2, 3, 4, 7 or 9, wherein if the moving object belongs to a high altitude toss, the moving characteristics of the moving object are added to a training set of a neural network discrimination model to perform model iteration.
CN202011224115.0A 2020-11-05 2020-11-05 High-altitude throwing object distinguishing method Active CN112329627B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011224115.0A CN112329627B (en) 2020-11-05 2020-11-05 High-altitude throwing object distinguishing method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011224115.0A CN112329627B (en) 2020-11-05 2020-11-05 High-altitude throwing object distinguishing method

Publications (2)

Publication Number Publication Date
CN112329627A true CN112329627A (en) 2021-02-05
CN112329627B CN112329627B (en) 2024-02-09

Family

ID=74315896

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011224115.0A Active CN112329627B (en) 2020-11-05 2020-11-05 High-altitude throwing object distinguishing method

Country Status (1)

Country Link
CN (1) CN112329627B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114241012A (en) * 2022-02-23 2022-03-25 深圳市研超科技有限公司 High-altitude parabolic determination method and device
CN116597340A (en) * 2023-04-12 2023-08-15 深圳市明源云科技有限公司 High altitude parabolic position prediction method, electronic device and readable storage medium

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104601939A (en) * 2013-10-31 2015-05-06 深圳市东宝嘉科技有限公司 Method and system for monitoring thrown object and eliminating risks of thrown object
CN110675592A (en) * 2019-08-16 2020-01-10 重庆特斯联智慧科技股份有限公司 High-altitude parabolic early warning protection system based on target identification and control method
CN111079663A (en) * 2019-12-19 2020-04-28 深圳云天励飞技术有限公司 High-altitude parabolic monitoring method and device, electronic equipment and storage medium
CN111553256A (en) * 2020-04-26 2020-08-18 上海天诚比集科技有限公司 High-altitude parabolic early warning identification method based on object track identification
CN111639578A (en) * 2020-05-25 2020-09-08 上海中通吉网络技术有限公司 Method, device, equipment and storage medium for intelligently identifying illegal parabola
CN111723654A (en) * 2020-05-12 2020-09-29 中国电子***技术有限公司 High-altitude parabolic detection method and device based on background modeling, YOLOv3 and self-optimization

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104601939A (en) * 2013-10-31 2015-05-06 深圳市东宝嘉科技有限公司 Method and system for monitoring thrown object and eliminating risks of thrown object
CN110675592A (en) * 2019-08-16 2020-01-10 重庆特斯联智慧科技股份有限公司 High-altitude parabolic early warning protection system based on target identification and control method
CN111079663A (en) * 2019-12-19 2020-04-28 深圳云天励飞技术有限公司 High-altitude parabolic monitoring method and device, electronic equipment and storage medium
CN111553256A (en) * 2020-04-26 2020-08-18 上海天诚比集科技有限公司 High-altitude parabolic early warning identification method based on object track identification
CN111723654A (en) * 2020-05-12 2020-09-29 中国电子***技术有限公司 High-altitude parabolic detection method and device based on background modeling, YOLOv3 and self-optimization
CN111639578A (en) * 2020-05-25 2020-09-08 上海中通吉网络技术有限公司 Method, device, equipment and storage medium for intelligently identifying illegal parabola

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
杨晔 等: "拉格朗日粒子模型在核事故应急中的开发与应用", 《原子能科学技术》, vol. 47, pages 713 - 714 *
赵璐 等: "基于多层感知器和运动轨迹的海上目标类型识别", 《舰船电子工程》, vol. 40, no. 3, pages 36 - 38 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114241012A (en) * 2022-02-23 2022-03-25 深圳市研超科技有限公司 High-altitude parabolic determination method and device
CN116597340A (en) * 2023-04-12 2023-08-15 深圳市明源云科技有限公司 High altitude parabolic position prediction method, electronic device and readable storage medium
CN116597340B (en) * 2023-04-12 2023-10-10 深圳市明源云科技有限公司 High altitude parabolic position prediction method, electronic device and readable storage medium

Also Published As

Publication number Publication date
CN112329627B (en) 2024-02-09

Similar Documents

Publication Publication Date Title
Fallon et al. Efficient scene simulation for robust Monte Carlo localization using an RGB-D camera
CN110889324A (en) Thermal infrared image target identification method based on YOLO V3 terminal-oriented guidance
CN112329627B (en) High-altitude throwing object distinguishing method
CN108256411A (en) By the method and system of camera review vehicle location
CN108399378B (en) Natural scene image identification method based on VGG deep convolution network
CN105550484A (en) Dynamic data driven modeling and simulation method for crowd evacuation and transportation behaviors in sudden-onset disaster
CN103778441B (en) A kind of sequence Aircraft Target Recognition based on DSmT and HMM
Kim et al. Convolutional neural network for monocular vision-based multi-target tracking
Mansour et al. Automated vehicle detection in satellite images using deep learning
CN106022266A (en) Target tracking method and target tracking apparatus
CN111415420B (en) Spatial information determining method and device and electronic equipment
Li et al. Intelligent mobile drone system based on real-time object detection
CN106127125A (en) Distributed DTW human body behavior intension recognizing method based on human body behavior characteristics
CN104102835A (en) Throw training score detection method based on real-time video images
CN108564600A (en) Moving object attitude tracking method and device
CN112347915B (en) High-altitude throwing object distinguishing system
CN105069829B (en) A kind of human body animation generation method based on more visually frequencies
CN114387462A (en) Dynamic environment sensing method based on binocular camera
Patil et al. A survey on joint object detection and pose estimation using monocular vision
Chen et al. Continuous occupancy mapping in dynamic environments using particles
Vrba et al. On onboard lidar-based flying object detection
CN106127119A (en) Joint probabilistic data association method based on coloured image and depth image multiple features
CN111414720A (en) Flow field vortex detection method based on neural network
Ozaki et al. Ekf-based real-time self-attitude estimation with camera dnn learning landscape regularities
CN115098827A (en) Infrared scene numerical value calculation method in snow accumulation environment

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