CN108470146B - Similar track identification method of classic track - Google Patents

Similar track identification method of classic track Download PDF

Info

Publication number
CN108470146B
CN108470146B CN201810143432.6A CN201810143432A CN108470146B CN 108470146 B CN108470146 B CN 108470146B CN 201810143432 A CN201810143432 A CN 201810143432A CN 108470146 B CN108470146 B CN 108470146B
Authority
CN
China
Prior art keywords
track
point
classical
real
compressed
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
CN201810143432.6A
Other languages
Chinese (zh)
Other versions
CN108470146A (en
Inventor
王前东
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.)
Southwest Electronic Technology Institute No 10 Institute of Cetc
Original Assignee
Southwest Electronic Technology Institute No 10 Institute of Cetc
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 Southwest Electronic Technology Institute No 10 Institute of Cetc filed Critical Southwest Electronic Technology Institute No 10 Institute of Cetc
Priority to CN201810143432.6A priority Critical patent/CN108470146B/en
Publication of CN108470146A publication Critical patent/CN108470146A/en
Application granted granted Critical
Publication of CN108470146B publication Critical patent/CN108470146B/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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/285—Selection of pattern recognition techniques, e.g. of classifiers in a multi-classifier system

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Navigation (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a method for identifying a classical track by using a phase recognition method, and aims to provide a classical track identification method which is high in similar track identification rate and can process unstable tracks. The invention is realized by the following technical scheme: reading a classical flight path from a classical flight path knowledge base, reading a real-time flight path from a real-time flight path base, compressing the real-time flight path by adopting a Douglas-Puck algorithm, performing initial flight path similarity judgment by utilizing flight path characteristics, if the initial judgment is successful, calculating the longest common substring distance of a plurality of pairs 1 by utilizing the distance between a point of the classical flight path and a segment of the real-time flight path, taking the longest common substring distance of the plurality of pairs 1 as the longest common substring distance of the plurality of pairs 1 between the classical flight path and the real-time flight path, taking the ratio of the longest common distance of the plurality of pairs 1 between the point and the line and the length of the classical flight path as flight path similarity, then performing flight path similarity accurate judgment according to the flight path similarity, and outputting a result if the flight path similarity accurate judgment is successful.

Description

Similar track identification method of classic track
Technical Field
The invention belongs to the field of pattern recognition, and relates to a classic track recognition technology in the field of intelligent information and the field of big data of the information.
Background
The classical flight path is the motion track of a classical target in the technical intelligence field. The activities of the classical target going out at each time are regular, and the motion trail is stable. The typical motion trajectory of such a classical target is a classical trajectory. The classical flight path plays a very important role in analysis of target identification, target warning, target behavior intention and the like. The similar track identification of the classic track is to identify the similar track of the real-time track in the classic track library. In practical situations, the acquired real-time flight path is very unstable, mainly reflected by:
1) the obtained track is discontinuous, is easy to miss and break, and forms an incomplete track;
2) the acquired flight path has large error and strong noise characteristic, and a strong noise flight path is formed;
3) the time delay of each point in the acquired track is not necessarily the same, the time interval of the track point has randomness, and an asynchronous unequal periodic track is formed.
Both the euclidean distance and the dynamic time warping distance are not suitable for such unstable tracks (incomplete tracks, strong noise tracks, asynchronous non-equal periodic tracks) due to the presence of incomplete tracks and strong noise tracks. The longest common substring distance can solve the recognition of incomplete tracks and strong noise tracks, but cannot directly process asynchronous unequal period tracks. When the longest common substring distance is applied to asynchronous unequal cycle step track recognition, three main problems need to be solved:
1) the method comprises the following steps of obtaining the problem that the period of a real-time track point is inconsistent with the period of a classic track point, wherein the distance difference between the real-time track point and the classic track point is large due to the fact that the periods of the real-time track point and the classic track point are inconsistent, the comparison result between the points is applied to the longest common substring distance algorithm, and the calculated track similarity is low and is inconsistent with the reality;
2) the real-time flight path and the classic flight path are not equal, and the classic flight path is formed by analyzing and summarizing the long-term activity rule, so that the classic flight path is more complete and accurate and has no strong noise compared with the real-time flight path. The plurality of points in the classical flight path may be on the same segment of the real-time flight path. The classical longest common substring distance is a distance of 1 to 1, and the distance of more than 1 is not considered, which is inconsistent with the actual requirement;
3) the calculation of the longest common substring distance is time-consuming, and an optimization algorithm is needed to reduce the calculation amount.
At present, when the longest common substring distance is applied to asynchronous unequal period track identification, the main solution is to convert the comparison between points into the comparison between line segments and line segments, and although the problem that the sampling period of a real-time track and a classic track is not synchronous is solved, the problem that a plurality of line segments of the classic track or one line segment of the real-time track are matched due to unequal classic track and real-time track cannot be solved, and the time consumption of the longest common substring distance when the longest common substring distance is used for track identification is not solved; in addition, a solution for track time consumption is provided, in the method, a plane where points are located forms a multilayer grid, each grid is replaced by one character, a multilayer character string recognition structure is formed, recognition is fast, but the method cannot solve recognition of asynchronous non-uniform period tracks.
Disclosure of Invention
The invention aims to solve the problems in the prior art and provides a rapid and robust classical track recognition method which has high similarity of similar tracks, high similar track recognition rate and good robustness of a calculation method, can process unstable tracks, improves the track recognition rate, reduces track recognition time and is high in speed.
In order to achieve the purpose, the invention provides a similar track identification method of a classical track, which is characterized by comprising the following steps: the classical track recognition module reads a classical track from a classical track knowledge base or establishes a rapid robust recognition model of a similar track by using a minimum rectangular area containing the classical track as a classical track characteristic; the classical flight path recognition module reads a real-time flight path from a real-time flight path library or uses a minimum rectangular area containing the real-time flight path as flight path characteristics of the real-time flight path, utilizes a distance algorithm from a classical flight path point to a real-time flight path segment and cooperates with a longest common substring distance algorithm of a plurality of pairs 1 to process an unstable asynchronous unequal periodic flight path, or adopts a Douglas-Peucker algorithm to compress the real-time flight path or utilizes the flight path characteristics to carry out initial flight path similarity judgment, if the initial flight path similarity judgment is successful, the distance between the points of the classical flight path and the segments of the real-time flight path is utilized to calculate the improved longest common substring distance of the plurality of pairs 1, the calculated longest common substring distance is used as the longest common substring distance between the points of the classical flight path and the real-time flight path and the longest common substring distance between the points of the plurality of pairs 1 and the lines, and the ratio of the length of the classical flight path is used as the similarity, and then, carrying out accurate track similarity judgment according to the track similarity, and outputting an identification result if the accurate track similarity judgment is successful.
Compared with the prior art, the invention has the following beneficial effects:
the similarity of similar tracks is high, the recognition rate of the similar tracks is high, and the robustness of the calculation method is good. The method reads the classical track from the classical track knowledge base or extracts the classical track characteristics, establishes a fast robust identification model of the similar track, utilizes the distance algorithm from the classical track point to the real-time track segment and matches with the longest common substring distance algorithm of a plurality of pairs 1, can process unstable asynchronous unequal periodic tracks, improves the track similarity and further improves the similar track identification rate.
The similar track recognition speed is high. The method adopts the Douglas-Peucker algorithm to compress the real-time track, improves the identification speed, reduces the number of the real-time track points after compression, and keeps the motion trend of the real-time track unchanged. The method adopts a point-to-line multi-pair 1 longest common substring distance algorithm to ensure that the track recognition rate is unchanged; the initial track similarity judgment with smaller calculated amount is used before the accurate track similarity judgment, so that the use times of the accurate track similarity judgment which takes more time are reduced.
The method can process unstable tracks, improve the track recognition rate and reduce the track recognition time. The method adopts the Douglas-Peucker algorithm to compress the real-time track, utilizes the distance between the point of the classical track and the segment of the real-time track and cooperates with the longest common substring distance algorithm of a plurality of pairs 1 to calculate the longest common substring distance between the point of the classical track and the real-time track and the longest common substring distance of the plurality of pairs 1 of the line, uses the improved ratio of the longest common substring distance between the point of the classical track and the line and the length of the classical track as the track similarity, and then carries out the track similarity precise judgment according to the track similarity, so that the unstable track can be processed, the track recognition rate is improved, and the track recognition time is reduced.
The method can be used for the application of activity rule analysis, moving target auxiliary identification and the like of moving targets such as airplanes, ships, automobiles, people and the like, can also be used for the cluster analysis of curves such as stocks, electrocardiograms and the like, and has strong engineering practical value.
Description of the drawings:
for a more clear understanding of the present invention, the invention will now be described by way of specific embodiments, with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of the recognition track identification of classical tracks of the present invention.
FIG. 2 is a schematic diagram of the track compression process of the Douglas-Peucker algorithm.
FIG. 3 is a trace plot of 10 (20) similar tracks tested in the examples.
FIG. 4 is a trajectory graph of 10 real-time tracks out of 10 pairs of similar tracks.
Fig. 5 is a trajectory diagram of 10 classical trajectories out of 10 pairs of similar trajectories.
Fig. 6 is a similarity comparison graph of 10 algorithms for similar tracks.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
See fig. 1. According to the invention, the following steps are adopted:
and S01, reading the classical track from the classical track knowledge base by the classical track recognition module.
And S02, the classical track recognition module extracts the classical track characteristics by using the minimum rectangular area containing the classical track as the track characteristics of the classical track [ optional step ], and the classical track characteristics are expressed by two rectangular vertex coordinates of the lower left corner and the upper right corner of the rectangle.
And S03, reading the real-time track from the real-time track database by the classical track recognition module.
And S04, extracting real-time track characteristics [ optional steps ], wherein the classical track recognition module uses a minimum rectangular area containing the real-time track as the track characteristics of the real-time track, and the classical track characteristics are expressed by coordinates of two rectangular vertexes of the lower left corner and the upper right corner of the rectangle.
S05, the classic track identification module adopts Douglas-Peucker algorithm to compress the real-time track or compress the real-time track [ optional step ].
S06, the classical track recognition module circularly traverses the classical track, and initial track similarity judgment is carried out by utilizing the track characteristics [ optional step ]; calculating the length of the x-axis side of the public rectangular area, and recording the length as xL; setting the length of the side length of the x axis of the characteristic of the classical track rectangular region as xcL, setting the initial judgment threshold value of the track similarity as eMT, when the xL is more than xcL × eMT, successfully judging the track similarity, ending the initial judgment process of the track similarity, and otherwise, performing the subsequent steps, calculating the side length of the y axis of the public rectangular region, and recording the length as yL; and (3) setting the length of the side length of the y axis of the rectangular region characteristic of the classical track as ycL, setting the initial judgment threshold value of the track similarity as eMT, when yL is more than ycL × eMT, the initial judgment of the track similarity is successful, otherwise, the initial judgment of the track similarity fails.
And S07, if the initial judgment of the similar tracks is successful, performing the subsequent steps, otherwise, turning to the step S06.
And S08, calculating the longest Common Substring distance by the classical track recognition module according to the distance between the point of the classical track and the segment of the real-time track, and calculating the longest Common Substring distance between the point of the classical track and the line of the real-time track by matching with an improved Long Common Substring distance algorithm (LCS) of multiple pairs 1.
S09, the classical track recognition module uses the ratio of the longest common substring distance calculated in the step S08 to the length of the classical track as the track similarity between the classical track and the real-time track.
S10, the classical track recognition module carries out track similarity precise judgment by utilizing the track similarity, the threshold value of the track similarity is eSimilar, the track similarity between the classical track and the real-time track calculated in the step S09 is f, when f is greater than eSimilar, the track similarity precise judgment between the classical track and the real-time track is successful, otherwise, the track similarity precise judgment between the classical track and the real-time track is failed.
And S11, if the accurate judgment of the similar tracks is successful, performing the subsequent steps, otherwise, turning to the step S06.
And S12, outputting the recognition result by the classical track recognition module, and ending the whole recognition process.
In step S01, assume that there are n classical trajectories { TC ] in the classical trajectory knowledge base1,TC2,…,TCnN is more than or equal to 1, the ith classical track TCiHas a position coordinate of (x)j,yj)，j＝1,2，…，ni，niCounting the number of the ith classic track, and generally requiring the number of the track to ensure the identification effectni≥5，i＝1,2，…n。
In step S02, the classical trajectory recognition module uses the coordinates of the bottom left corner and the top right corner of the minimum rectangle in which the classical trajectory is located to represent the trajectory rectangular region feature, and assumes that the rectangular region features of n classical trajectories are MCiThe rectangular area features of (xMin, yMin, xMax, yMax), i ═ 1,2, …, n, the ith classical track can be obtained by equations (1) to (4).
Figure BDA0001578197270000041
Figure BDA0001578197270000042
Figure BDA0001578197270000043
Figure BDA0001578197270000044
Wherein MC isi(xMin, yMin, xMax, yMax) is a rectangular feature of the ith classical track, MCixMin is the x-axis coordinate value of the lower left corner of the rectangular feature, MCiyMin is the y-axis coordinate value of the lower left corner of the rectangular feature, MCixMax is the x-axis coordinate value of the top right corner of the rectangular feature, MCiyMax is the y-axis coordinate value of the upper right corner of the rectangle; TC (tungsten carbide)iIs the ith classic track, TCi.x1Is the x-axis coordinate value, TC, of the 1 st point of the ith classical tracki.x2Is the x-axis coordinate value of the 2 nd point of the ith classical track,
Figure BDA0001578197270000045
is the ith classic trackniX-axis coordinate value of point, TCi.y1Is the y-axis coordinate value, TC, of the 1 st point of the ith classical tracki.y2Is the y-axis coordinate value of the 2 nd point of the ith classical track,
Figure BDA0001578197270000046
is the ith classic trackniThe y-axis coordinate value of the point, min represents the function to find the minimum value in the set, and max represents the function to find the maximum value in the set.
In step S03, the real-time track read by the classical track recognition module is TR, and the position coordinate of the real-time track is (x)j,yj)，j＝1,2，…，m1，m1The number of track points of the real-time track TR.
In step S04, the classical flight path recognition module represents the flight path rectangular area feature by using the lower left-hand coordinate and the upper right-hand coordinate of the minimum rectangle in which the real-time flight path TR is located, and the real-time flight path rectangular area feature can be obtained by equations (5) to (8).
Figure BDA0001578197270000051
Figure BDA0001578197270000052
Figure BDA0001578197270000053
Figure BDA0001578197270000054
Wherein MR (xMin, yMin, xMax, yMax) is a rectangular feature of the real-time track, MR.xMin is an x-axis coordinate value of the lower left corner of the rectangular feature, MR.yMin is a y-axis coordinate value of the lower left corner of the rectangular feature, MR.xMax is an x-axis coordinate value of the upper right corner of the rectangular feature, and MR.yMax is a y-axis coordinate value of the upper right corner of the rectangular feature; TR.x1X-axis coordinate value of point 1 of real-time track TR, TR.x2Is the x-axis coordinate value of the 2 nd point of the real-time track TR,
Figure BDA0001578197270000055
mth for real-time track1X-axis of pointStandard value, TR.y1Y-coordinate value of point 1 of real-time track TR, TR2The y-axis coordinate value of the 2 nd point of the compressed track TR,
Figure BDA0001578197270000056
mth for real-time track1Y-axis coordinate value of point, m1And the number of the track points of the real-time track TR is min, which represents a function for solving the minimum value in the set, and max, which represents a function for solving the maximum value in the set.
In step S05, the algorithm used for track compression is Douglas-pocklac Douglas-Peucker algorithm, and the algorithm is implemented by the following steps:
s05.01, the classic track identification module takes the real-time track TR as the track DTR to be compressed, and the number m of the track points of the DTR is equal to the number m of the track points of the TR1Initializing a compressed track CTR as an empty set, subsequently sequentially judging whether points in a track DTR to be compressed are reserved, and if so, adding the points into the compressed track CTR.
S05.02, calculating all points P in the DTR of the flight path to be compressed by a classical flight path identification modulejTo a line segment formed by head and tail of a flight path to be compressed
Figure BDA0001578197270000057
The distance between
Figure BDA0001578197270000058
j is 1,2, …, m, let x1Is a point P1X-axis coordinate value of (2), y1Is a point P1Y-axis coordinate value of (2), xmIs a point PmX-axis coordinate value of (2), ymIs a point PmY-axis coordinate value of (2), xjIs a point PjX-axis coordinate value of (2), yjIs a point PjY-axis coordinate value of (1), then
Figure BDA0001578197270000059
The calculation steps are as follows:
s05.02.01, calculating line segment
Figure BDA00015781972700000510
Represented position coordinate vector and line segment
Figure BDA00015781972700000511
The inner product f, f of the expressed position coordinate vector is obtained by equation (9).
f＝(xj-x1)×(xm-x1)+(yj-y1)×(ym-y1)(9)
S05.02.02, if f is less than or equal to 0, then point PjTo line segment
Figure BDA00015781972700000512
Is a distance of
Figure BDA00015781972700000513
This can be determined by equation (10) and step S05.02 ends, otherwise step S05.02.03 is executed.
Figure BDA0001578197270000061
S05.02.03, calculating line segment
Figure BDA0001578197270000062
D is the modulus of the position coordinate vector expressed by equation (11).
d＝(xm-x1)2+(ym-y1)2(11)
S05.02.04, if f-d is not less than 0 or d is not more than 0, the point PjTo line segment
Figure BDA0001578197270000063
Is a distance of
Figure BDA0001578197270000064
Can be found by the formula (12),
Figure BDA0001578197270000065
step S05.02 ends, otherwise step S05.02.05 is executed.
S05.02.05, calculating a point PjTo line segment
Figure BDA0001578197270000066
The vertical angle of (d); let the vertical angle be PsThe vertical angle coordinate position is (x)s，ys) Then xsAnd ysThis can be obtained by the equations (13) and (14).
Figure BDA0001578197270000067
Figure BDA0001578197270000068
S05.02.06, angle P determined at step S05.02.05sPoint PjTo line segment
Figure BDA0001578197270000069
Is a distance of
Figure BDA00015781972700000610
Can be found by the formula (15),
Figure BDA00015781972700000611
step S05.02 ends.
S05.03, the classical track recognition module compares the m real-time track points P calculated in the step S05.02jSegment composed of head and tail of real-time track
Figure BDA00015781972700000612
M distances of
Figure BDA00015781972700000613
J is 1,2, …, m, the point with the largest distance is selected, and the selected j-th point is made1The point is the point with the largest distance, then the maximum distance is
Figure BDA00015781972700000614
I.e. j1For all j ═ 1,2, …, m is given by formula (16).
Figure BDA00015781972700000615
S05.04, the classical track identification module calculates the maximum distance according to the step S05.03
Figure BDA00015781972700000616
J is judged1Whether a point is kept in the compressed track CTR. If it is not
Figure BDA00015781972700000617
If the track compression threshold value eCompressDis is not greater than the track compression threshold value eCompressDis, no point in the track to be compressed DTR is added into the compressed track CTR, and the track compression process of the track to be compressed DTR is finished. If it is not
Figure BDA00015781972700000618
And if the track compression threshold value is larger than the eCommressDis, executing the subsequent steps.
S05.05, the classical track identification module carries out the 1 st point to the jth point1The first section of the track of the point is used as a new track DTR to be compressed, and the new track to be compressed has j1A track point, let m be j1The new track to be compressed DTR is compressed starting from step S05.02.
S05.06, the classic track identification module will j1The point is added from the tail to the compressed track CTR as a reserved point.
S05.07, the classic track identification module will j1Taking a second section of track from the point to the mth point as a new track to be compressed DTR, wherein the track to be compressed has m-j1+1 track points, let m equal m-j1+1, the new track to be compressed DTR is compressed starting from step S05.02.
S05.08, adding the 1 st point into a track head of a compressed track CTR by a classic track identification module, and adding the m th point into the track head of the compressed track CTR1And adding the points into the track tail of the compressed track CTR, wherein the compressed track CTR is the compressed track of the real-time track.
See fig. 2. The flight path indicated by the dashed line is the real-time flight path TR to be compressed. Among all the points from "star" to "end", a point "1" is farthest from the line segment "star" and "end", the farthest distance is greater than a threshold value eCompressDis, the point "1" is reserved, the point "1" will be added into the compressed track CTR, and the point "1" divides the track into two sections: "star" to "1" and "1" to "end".
In the track segment from "star" to "1", the point "2" is farthest from the segment "star" and "1", and the farthest distance is greater than the threshold value eCompressDis, so that the point "2" is reserved, and the point "2" will be added to the compressed track CTR. The point "2" divides the track segment "star" to "1" into two segments: "star" to "2" and "2" to "1".
In the track segment from '1' to 'end', a point '3' is farthest away from the segment '1' and 'end', and the farthest distance is greater than a threshold value eCompressDis, so that the point '3' is reserved, and the point '3' is added into a compressed track CTR. The point 3 divides the track section from 1 to end into two sections: "1" to "3" and "3" to "end".
So far, the point "1", the point "2", the point "3" divides the track into four segments of "star" to "2", "2" to "1", "1" to "2", and "3" to "end", and in the segment "star" to "2", no track point is more than the threshold value eCompressDis from the segment "star" to the segment "star" 2 ", so no track point is reserved in the segment" star "to" 2 "; also, no waypoint is retained in the three segments "2" to "1", "1" to "2", and "3" to "end".
After the classical track recognition module adds the head and tail track points to the compressed track, the compressed track CTR has 5 points including "star", "2", "1", "3" and "end", and the track represented by the solid line in FIG. 2 is the compressed track CTR.
In step S08, the classical trajectory recognition module calculates the classical trajectory and the real-time trajectory by using the improved longest Common Substring distance algorithm (LCS) of the point-to-line multiple pairs 1The longest common substring distance between the point and the line of the multiple pairs 1 is set as the ith classical track
Figure BDA0001578197270000071
Wherein (x)1,y1)、
(x2,y2)、…、
Figure BDA0001578197270000072
Respectively classic track TCiPoint 1, point 2, …, point 1niSetting the coordinate value of point and the real-time track after compression as the compressed track
Figure BDA0001578197270000073
(x1,y1)、(x2,y2)、…、
Figure BDA0001578197270000074
Respectively, the 1 st point, the 2 nd point, … th point and the second point of the compressed track CTRm2If the coordinate value of the point and the Euclidean distance threshold are eDis, the implementation mode of the longest common substring distance algorithm of multiple pairs 1 of point-to-line between the classical flight path and the real-time flight path comprises the following steps:
s08.01, let LeDisIs a classic track TCiAnd a length matrix of the compressed track CTR, the size of the matrix isni+1)*
(m2+1), initialize the 0 th row element and 0 column element of the matrix with 0.
S08.02, calculating the jth point C of the classical trackjThe coordinate value is (TC)i.xj,TCi.yj) To the kth line segment of the compressed track
Figure BDA0001578197270000081
The distance between
Figure BDA0001578197270000082
The kth line segment
Figure BDA0001578197270000083
The end point of (1) is the kth point R of the compressed flight pathkAnd the (k +1) th point Rk+1Point RkIs (CTR.x)k,CTR.yk) Point Rk+1Has a coordinate value of (CTR.x)k+1,CTR.yk+1)，
Figure BDA0001578197270000084
The calculation process of (2) includes:
s08.02.01, calculating line segment
Figure BDA0001578197270000085
Represented position coordinate vector and line segment
Figure BDA0001578197270000086
The inner product f of the represented position coordinate vectors,
f is obtained by equation (17).
f＝(TCi.xj-CTR.xk)×(CTR.xk+1-CTR.xk)+(TCi.yj-CTR.yk)×(CTR.yk+1-CTR.yk) (17) S08.02.02, if f is less than or equal to 0, the point CjTo line segment
Figure BDA0001578197270000087
Is a distance of
Figure BDA0001578197270000088
Can be found by equation (18) and step S08.02 ends, otherwise step S08.02.03 is performed.
Figure BDA0001578197270000089
S08.02.03, calculating line segment
Figure BDA00015781972700000810
D is the modulus of the position coordinate vector expressed by equation (19).
d＝(CTR.xk+1-CTR.xk)2+(CTR.yk+1-CTR.yk)2(19) S08.02.04, if f-d is not less than 0 or d is not more than 0, the point CjTo line segment
Figure BDA00015781972700000811
Is a distance of
Figure BDA00015781972700000812
Can be found by equation (20) and step S08.02 ends, otherwise step S08.02.05 is performed.
Figure BDA00015781972700000813
S08.02.05, calculate Point CjTo line segment
Figure BDA00015781972700000814
The vertical angle of (d); let the vertical angle be PsThe coordinate position is (x)s，ys) Then xsAnd ysThis can be obtained by the formula (21) and the formula (22).
Figure BDA00015781972700000815
Figure BDA00015781972700000816
S08.02.06, angle P determined at step S08.02.05sPoint CjTo line segment
Figure BDA00015781972700000817
Is a distance of
Figure BDA00015781972700000818
This can be obtained by the formula (23), and step S08.02 ends.
Figure BDA00015781972700000819
S08.03, the classic track identification module calculates the TC of the classic track by using a recursion formula (24) according to the distance between the point of the classic track calculated in the step S08.02 and the segment of the compressed track (the compressed real-time track)iAnd length matrix L of compressed flight path CTReDisThen, the classical flight path TC is obtainediMultiple pairs 1 of point-to-line longest common substring distance l between the compressed flight path CTRi＝LeDis(ni,m2-1)，
Figure BDA0001578197270000091
Wherein eDis is Euclidean distance threshold, CjIs a classic track TCiAt the point of the (j) th,
Figure BDA0001578197270000092
to compress the kth segment of the track CTR,
Figure BDA0001578197270000093
point C calculated for step S08.02jAnd line segment
Figure BDA0001578197270000094
Between Euclidean distance, max represents a function taking the maximum value in the set, LeDis(j, k-1) represents a length matrix LeDisJ (th) row and k-1 column of elements, LeDis(j-1, k) represents a length matrix LeDisJ-1 row, k column element, LeDis(j, k) represents a length matrix LeDisRow j and column k.
In step S09, let classical track TCiThe longest common substring distance of the pairs 1 of point-to-line between the compressed tracks CTR is liClassic track TCiLength niThen classic track TCiSimilarity with the compressed track CTR is fi， fiCan be obtained by the formula (25).
fi＝li/ni(25)
In the specific implementation step S08.03, the classical trajectory recognition module uses a new longest common substring distance formula (24) of multiple pairs 1 of point-to-line when calculating the longest common substring distance, and compared with the existing longest common substring distance formula (26) of pairs 1 of point-to-point 1, the method has two improvements: on one hand, formula (26) uses the Euclidean distance formula from point to point, and formula (24) uses the Euclidean distance formula from point to line segment; on the other hand, equation (26) uses the longest common substring distance equation of 1 to 1, and equation (24) uses the longest common substring distance equation of more than one pair 1. To illustrate the necessity of performing both improvements, it is proposed to improve only one of the equations (27) and (28).
In the specific implementation step S08.03, the classical trajectory recognition module calculates the classical trajectory TC using the classical point-to-point 1-to-1 longest common substring distance formula (26)iLength matrix L1 with compressed track CTReDisThen, the classical flight path TC is obtainediThe longest distance l between the point-to-point 1 and the 1 common substring between the compressed flight path CTR and the point-to-point 1i＝L1eDis(ni,m2)，
Figure BDA0001578197270000095
Wherein eDis is Euclidean distance threshold, CjIs a classic track TCiJ point of (1), RkTo compress the kth point of the track CTR, dis (C)j,Rk) Is point CjAnd point RkBetween Euclidean distance, max represents a function taking the maximum value in the set, L1eDis(j, k-1) denotes a length matrix L1eDisJ (th) row (k-1) column element, L1eDis(j-1, k) denotes a length matrix L1eDisLine j-1, column k element of L1eDis(j-1, k-1) represents the length matrix L1eDisLine j-1, column k-1 element of L1eDis(j, k) denotes a length matrix L1eDisRow j and column k.
In step S08.03, the classical trajectory recognition module uses the improved longest common substring distance formula from point to point in multiple pairs of 1(26) Calculating the classic track TCiLength matrix L2 with compressed track CTReDisThen, the classical flight path TC is obtainediPoint-to-point multiple-1 pairs of longest common substring distances l from compressed flight path CTRi＝L2eDis(ni,m2)，
Figure BDA0001578197270000101
Wherein eDis is Euclidean distance threshold, CjIs a classic track TCiJ point of (1), RkTo compress the kth point of the track CTR, dis (C)j,Rk) Is point CjAnd point RkBetween Euclidean distance, max represents a function taking the maximum value in the set, L2eDis(j, k-1) denotes a length matrix L2eDisJ (th) row (k-1) column element, L2eDis(j-1, k) denotes a length matrix L2eDisLine j-1, column k element of (L2)eDis(j, k) denotes a length matrix L2eDisRow j and column k.
Specifically, in step S08.03, the classic track identification module calculates the classic track TC using the improved point-to-line 1-to-1 recursion formula (28) according to the distance between the point of the classic track calculated in step S08.02 and the segment of the compressed track (compressed real-time track)iLength matrix L3 with compressed track CTReDisThen, the classical flight path TC is obtainediThe longest common substring distance l from point to line 1 to 1 between the compressed flight path CTRi＝L3eDis(ni,m2-1),
Figure BDA0001578197270000102
Wherein eDis is Euclidean distance threshold, CjIs a classic track TCiAt the point of the (j) th,
Figure BDA0001578197270000103
to compress the kth segment of the track CTR,
Figure BDA0001578197270000104
point C calculated for step S08.02jAnd line segment
Figure BDA0001578197270000105
Between Euclidean distance, max represents a function taking the maximum value in the set, L3eDis(j, k-1) represents a length matrix L3eDisJ (th) row (k-1) column element, L3eDis(j-1, k) denotes a length matrix L3eDisLine j-1, column k element of L3eDis(j-1, k-1) represents the length matrix L3eDisLine j-1, column k-1 element of L3eDis(j, k) denotes a length matrix L3eDisRow j and column k.
Examples
See fig. 3. The path similar rectangular area has 20 test target paths, 10 of which are marked as paths: target 1, target 3, …, and target 19 are used as real-time tracks, and are referred to as TR ═ TR1, TR2, …, TR10], as shown in fig. 4; simultaneously, 10 track targets: target 2, target 4, …, and target 20 are given as a classic track as TC ═ TC1, TC2, …, and TC10, as shown in fig. 5.
Ignoring the reading of data, assume that 10 classical flight paths LisTC as shown in fig. 5 [ TC1, TC2, …, TC10] have been read in the memory, and 10 real-time flight paths LisTR as shown in fig. 4 [ TR1, TR2, …, TR10] have been read. The similarity calculation method in total 8 calculates the track similarity between the real-time track and the classical track according to whether the optional implementation step of track compression S05 is adopted and the formula adopted in the implementation step S08.03 is different from the formula (26), the formula (27), the formula (28) or the formula (24): the 1 st similarity calculation method does not adopt the optional implementation step S05, namely, does not compress the real-time track, and adopts the longest common substring distance formula (26) of point-to-point 1 to 1 in the implementation step S08.03; the 2 nd similarity calculation method does not adopt the optional implementation step S05, namely, does not compress the real-time track, and adopts the longest common substring distance formula (27) of point-to-point multiple pairs 1 in the implementation step S08.03; the similarity calculation method 3 does not adopt the optional implementation step S05, namely, does not compress the real-time flight path, and adopts the longest common substring distance formula (28) of point-to-line 1 to 1 in the implementation step S08.03; the 4 th similarity calculation method does not adopt the optional implementation step S05, namely, does not compress the real-time flight path, and adopts the longest common substring distance formula (24) of multiple point-to-line pairs 1 in the implementation step S08.03; the 5 th similarity calculation method adopts an optional implementation step S05, namely compressing the real-time track, and adopts a longest common substring distance formula (26) of point-to-point 1 to 1 in an implementation step S08.03; the similarity calculation method of the 6 th adopts an optional implementation step S05, namely compressing the real-time track, and in the implementation step S08.03, adopts a longest common substring distance formula (27) of point-to-point multiple pairs 1; the 7 th similarity calculation method adopts an optional implementation step S05, namely compressing the real-time flight path, and adopts a longest common substring distance formula (28) of point-to-line 1 to 1 in an implementation step S08.03; the 8 th similarity calculation method adopts an optional implementation step S05, namely compressing the real-time flight path, and in an implementation step S08.03, adopts a longest common substring distance formula (24) of multiple pairs 1 of point-to-line, which is also the method used in the present patent. These 8 similarity calculation methods are respectively named as: "Point-to-Point 1 vs. 1"; "Point-to-point multiple pairs of 1"; "Point-to-line 1 vs. 1"; "Point-to-line multiple pairs of 1"; "Point-to-Point 1 to 1 after compression"; "Point-to-point multiple pairs 1 after compression"; "Point to line 1 to 1 after compression"; "Point-to-line multiple pairs 1 after compression". "Point-to-line multiple pairs 1 after compression" is the method used in this patent.
The compression threshold eCompressDis set in the optional implementation step S05 is 10000 meters, the euclidean distance threshold eDis in the longest common substring distance algorithm is 10000 meters, the processor in the Win7 operating system, intel (r) CoreTMi5-6500CPU @3.20GHz, the track similarity calculated by the above 8 similarity calculation method on Matlab2009 software is M1, M2, …, M8, and the operation processing time is F [1], F [2], …, F [8] in F. The processing time of the run is shown in table 1. The similarity of the diagonal lines in the M1, M2, M … and M8 is the track similarity between the TC1, TC2, … and TC10 calculated by 8 algorithms and the corresponding similar tracks TR1, TR2, … and TR10, as shown in table 2, and the similarity of 10 pairs of similar tracks is plotted as shown in fig. 6.
TABLE 1
Figure BDA0001578197270000111
As can be seen from table 1, compressing the tracks can reduce the processing time by more than 90%, which illustrates that the track compression speeds up the processing speed of similar track identification.
TABLE 2 track similarity
Figure BDA0001578197270000121
As can be seen from table 2 and fig. 6:
similarity calculated by four algorithms (point-to-point 1 to 1, point-to-point multiple-to-1, point-to-line 1 to 1 and point-to-line multiple-to-1) of the real-time flight path is close, and the similarity is more than 0.5, so that similar flight paths can be identified;
the similarity calculated by the three algorithms (point to point 1, point to point multiple pairs 1 after compression, point to line 1 to 1 after compression) after real-time track compression is less than 0.3, which shows that the track compression reduces the track similarity calculated by the three algorithms (point to point 1, point to point multiple pairs 1, point to line 1 to 1), and shows that the three algorithms are not robust to the compressed track (incomplete, asynchronous unequal period).
The flight path similarity calculated by the algorithm of 'point-to-line multi-pair 1' after compression is close to the similarity calculated by four uncompressed algorithms (point-to-point 1-to-1, point-to-point multi-pair 1, point-to-line 1-to-1 and point-to-line multi-pair 1), but the processing time is much shorter than that of the four uncompressed algorithms, which shows that the algorithm of point-to-line multi-pair 1 has no influence on flight path compression, can solve the recognition of similar flight paths of incomplete asynchronous non-equal period data, and also shows that the algorithm of 'point-to-line multi-pair 1' after compression can accelerate the recognition processing speed.

Claims (9)

1. A similar flight path identification method of a classic flight path is characterized by comprising the following steps: the classical track recognition module reads a classical track from a classical track knowledge base or establishes a rapid robust recognition model of a similar track by using a minimum rectangular area containing the classical track as a classical track characteristic; the classic track recognition module reads real-time tracks from a real-time track library or uses a minimum rectangular area containing the real-time tracks as track characteristics of the real-time tracks, utilizes a distance algorithm from classic track points to real-time track segments and matches with a plurality of pairs 1 of longest common substring distance algorithms to process unstable asynchronous unequal periodic tracks, compresses the real-time tracks by adopting a Douglas-Peucker algorithm, performs initial track similarity judgment by utilizing the track characteristics, and if the initial track similarity judgment is successful, calculates the improved pairs 1 of longest common substring distances by utilizing the distance from the points of the classic tracks to the segments of the real-time tracks, and calculates by adopting the following calculation steps:
step 01, let LeDisIs a classic track TCiAnd a length matrix of the compressed track CTR, the size of the matrix isni+1)*(m2+1), initializing the 0 th row element and 0 column element of the matrix with 0;
step 02, calculating the jth point C of the classical flight pathjThe coordinate value is (TC)i.xj,TCi.yj) To the kth line segment of the compressed track
Figure FDA0003340011360000011
The distance between
Figure FDA0003340011360000012
The kth line segment
Figure FDA0003340011360000013
The end point of (1) is the kth point R of the compressed flight pathkAnd the (k +1) th point Rk+1Point RkHas a coordinate value of (CTR.x)k,CTR.yk) Point Rk+1Has a coordinate value of (CTR.x)k+1,CTR.yk+1)；
Step 03, the classical track identification module calculates the classical track TC by the following recursion formula according to the distance between the point of the classical track calculated in step 02 and the segment of the compressed real-time trackiLength matrix L with compressed track CTReDisThen, the classical flight path TC is obtainediAnd the longest common substring distance between the compressed flight path CTR and the point-to-line pairs 1: li＝LeDis(ni,m2-1),
Figure FDA0003340011360000014
The calculated longest common substring distance is used as the longest common substring distance of multiple pairs 1 of point-to-line between the classical track and the real-time track, the ratio of the longest common substring distance of the multiple pairs 1 of point-to-line and the length of the classical track is used as the track similarity, then the track similarity precise judgment is carried out according to the track similarity, if the track similarity precise judgment is successful, the identification result is output, wherein eDis is an Euclidean distance threshold, C is a Euclidean distance threshold, andjis a classic track TCiAt the point of the (j) th,
Figure FDA0003340011360000015
to compress the kth segment of the track CTR,
Figure FDA0003340011360000016
point C calculated for step 02jAnd line segment
Figure FDA0003340011360000017
Between Euclidean distance, max represents a function taking the maximum value in the set, LeDis(j, k-1) represents a length matrix LeDisJ (th) row and k-1 column of elements, LeDis(j-1, k) represents a length matrix LeDisJ-1 row, k column element, LeDis(j, k) represents a length matrix LeDisRow j and column k.
2. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: the classic track recognition module uses a minimum rectangular area containing the real-time track as a track characteristic of the real-time track, and the classic track characteristic is represented by coordinates of two rectangular vertexes of the lower left corner and the upper right corner of the rectangle.
3. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: and the classical track recognition module carries out track similarity precise judgment by utilizing the track similarity, the threshold value for judging the track similarity is eSimilar, the track similarity is f, when f is greater than eSimilar, the track similarity precise judgment of the classical track and the real-time track is successful, otherwise, the track similarity precise judgment of the classical track and the real-time track fails.
4. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: the classic track knowledge base comprises n classic tracks { TC1,TC2,…,TCnN is more than or equal to 1, the ith classical track TCiHas a position coordinate of (x)j,yj)，j＝1,2，…，ni，niIs the number of points of the ith classical track, and ni≥5，i＝1,2，…n。
5. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: the rectangular region of n classical tracks is characterized by MCi(xMin, yMin, xMax, yMax), i ═ 1,2, …, n, the rectangular area characteristic of the ith classical track is determined by the following formula,
Figure FDA0003340011360000021
Figure FDA0003340011360000022
Figure FDA0003340011360000023
Figure FDA0003340011360000024
wherein, MCi(xMin, yMin, xMax, yMax) is a rectangular feature of the ith classical track, MCixMin is the x-axis coordinate value of the lower left corner of the rectangular feature, MCiyMin is the y-axis coordinate value of the lower left corner of the rectangular feature, MCixMax is the x-axis coordinate value of the top right corner of the rectangular feature, MCiyMax is the y-axis coordinate value of the upper right corner of the rectangle; TC (tungsten carbide)iIs the ith classic track, TCi.x1Is the x-axis coordinate value, TC, of the 1 st point of the ith classical tracki.x2Is the x-axis coordinate value of the 2 nd point of the ith classical track,
Figure FDA0003340011360000025
is the nth of the ith classical trackiX-axis coordinate value of point, TCi.y1Is the y-axis coordinate value, TC, of the 1 st point of the ith classical tracki.y2Is the y-axis coordinate value of the 2 nd point of the ith classical track,
Figure FDA0003340011360000026
is the nth of the ith classical trackiThe y-axis coordinate value of the point, min represents the function to find the minimum value in the set, and max represents the function to find the maximum value in the set.
6. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: the coordinates of the lower left corner and the coordinates of the upper right corner of the minimum rectangle where the real-time track TR is positioned represent the characteristics of the track rectangular area, the characteristics of the real-time track rectangular area can be obtained by the following formula,
Figure FDA0003340011360000027
Figure FDA0003340011360000028
Figure FDA0003340011360000029
Figure FDA00033400113600000210
wherein MR (xMin, yMin, xMax, yMax) is a rectangular feature of the real-time track, MR.xMin is an x-axis coordinate value of the lower left corner of the rectangular feature, MR.yMin is a y-axis coordinate value of the lower left corner of the rectangular feature, MR.xMax is an x-axis coordinate value of the upper right corner of the rectangular feature, and MR.yMax is a y-axis coordinate value of the upper right corner of the rectangular feature; TR.x1X-axis coordinate value of point 1 of real-time track TR, TR.x2Is the x-axis coordinate value of the 2 nd point of the real-time track TR,
Figure FDA0003340011360000031
mth for real-time track1X-axis coordinate value of point, TR.y1Y-coordinate value of point 1 of real-time track TR, TR2The y-axis coordinate value of the 2 nd point of the compressed track TR,
Figure FDA0003340011360000032
mth for real-time track1Y-axis coordinate value of point, m1And the number of the track points of the real-time track TR is min represents a function for solving the minimum value in the set, and max represents a function for solving the maximum value in the set.
7. The method for identifying similar tracks in classic tracks according to claim 1, characterized in that: the method adopts a Douglas-Peucker algorithm to compress the real-time flight path, and the implementation steps of the algorithm comprise: step 01, the classic track identification module takes the real-time track TR as the pressure to be measuredThe contracted track DTR, the number m of the track points of DTR is equal to the number m of the track points of TR1Initializing a compressed track CTR as an empty set, subsequently sequentially judging whether points in a track DTR to be compressed are reserved, and if so, adding the points into the compressed track CTR; step 02, calculating all points P in the to-be-compressed track DTR by the classical track recognition modulejTo a line segment formed by head and tail of a flight path to be compressed
Figure FDA0003340011360000033
The distance between
Figure FDA0003340011360000034
Step 03, the classical track identification module compares the m real-time track points P calculated in step S05.02jSegment composed of head and tail of real-time track
Figure FDA0003340011360000035
M distances of
Figure FDA0003340011360000036
J is 1,2, …, m, the point with the largest distance is selected, and the selected j-th point is made1The point is the point with the largest distance; step 04, the classical track recognition module calculates the maximum distance according to step 03
Figure FDA0003340011360000037
J is judged1Whether or not a point is retained in the compressed track CTR, if
Figure FDA0003340011360000038
If the track compression threshold value eCommpressDis is not larger than the track compression threshold value eCommpressDis, no point in the track to be compressed DTR is added into the compressed track CTR, the track compression process of the track to be compressed DTR is ended, and if the track to be compressed DTR is not larger than the track compression threshold value eCommpressDis, the track compression process of the track to be compressed DTR is ended
Figure FDA0003340011360000039
If the track compression threshold value is larger than the eCommressDis, executing the subsequent steps; step 05, the classic track identification module will1 st to j1The first section of the track of the point is used as a new track DTR to be compressed, and the new track to be compressed has j1A track point, let m be j1Compressing the new track DTR to be compressed from the step 02; step 06, the classic track identification module will be the jth1Adding the points as retention points into the compressed flight path CTR from the tail part; step 07, the classic track identification module sends the jth1Taking a second section of track from the point to the mth point as a new track to be compressed DTR, wherein the track to be compressed has m-j1+1 track points, let m equal m-j1+1, compressing the new track to be compressed DTR from step 02; step 08, the classic track identification module adds the 1 st point to the track head of the compressed track CTR and adds the mth point to the track head1And adding the points into the flight path tail of the compressed flight path CTR, wherein the compressed flight path CTR is the compressed flight path of the real-time flight path.
8. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: the classic track identification module calculates all points P in the DTR of the track to be compressedjTo a line segment formed by head and tail of a flight path to be compressed
Figure FDA00033400113600000310
The distance between
Figure FDA00033400113600000311
Is calculated by let x1Is a point P1X-axis coordinate value of (2), y1Is a point P1Y-axis coordinate value of (2), xmIs a point PmX-axis coordinate value of (2), ymIs a point PmY-axis coordinate value of (2), xjIs a point PjX-axis coordinate value of (2), yjIs a point PjY-axis coordinate value of (1), then
Figure FDA0003340011360000041
The calculation comprises the following calculation steps:
step 01, calculating line segment
Figure FDA0003340011360000042
Represented position coordinate vector and line segment
Figure FDA0003340011360000043
The inner product f, f of the expressed position coordinate vector is obtained by the formula (9);
f＝(xj-x1)×(xm-x1)+(yj-y1)×(ym-y1) (9)
step 02, if f is less than or equal to 0, then point PjTo line segment
Figure FDA0003340011360000044
Is a distance of
Figure FDA0003340011360000045
The calculation is finished through the formula (10), otherwise, the step 03 is executed;
Figure FDA0003340011360000046
step 03, calculating line segments
Figure FDA0003340011360000047
D is the modulus d of the position coordinate vector expressed by the formula (11);
d＝(xm-x1)2+(ym-y1)2 (11)
step 04, if f-d is not less than 0 or d is not more than 0, then point PjTo line segment
Figure FDA0003340011360000048
Is a distance of
Figure FDA0003340011360000049
Can be obtained by the formula (12);
Figure FDA00033400113600000410
finishing the calculation, otherwise, executing the step 05;
step 05, calculate point PjTo line segment
Figure FDA00033400113600000411
The vertical angle of (d); let the vertical angle be PsThe vertical angle coordinate position is (x)s，ys) Then xsAnd ysCan be obtained by the formula (13) and the formula (14);
Figure FDA00033400113600000412
Figure FDA00033400113600000413
step 06, using the vertical angle P obtained in step 05sPoint PjTo line segment
Figure FDA00033400113600000414
Is a distance of
Figure FDA00033400113600000415
Can be obtained by the formula (15);
Figure FDA00033400113600000416
and finishing the calculation.
9. The method for identifying similar tracks in classical tracks according to claim 1, characterized in that: performing initial track similarity judgment by using track characteristics, and calculating the length of the x-axis side of the public rectangular area, wherein the length of the x-axis side is recorded as xL; setting the length of the side length of the x axis of the characteristic of the classical track rectangular region as xcL, setting the initial judgment threshold value of the track similarity as eMT, when the xL is more than xcL × eMT, successfully judging the track similarity, ending the initial judgment process of the track similarity, and otherwise, performing the subsequent steps, calculating the side length of the y axis of the public rectangular region, and recording the length as yL; and (3) setting the length of the side length of the y axis of the rectangular region characteristic of the classical track as ycL, setting the initial judgment threshold value of the track similarity as eMT, when yL is more than ycL × eMT, the initial judgment of the track similarity is successful, otherwise, the initial judgment of the track similarity fails.
CN201810143432.6A 2018-02-11 2018-02-11 Similar track identification method of classic track Active CN108470146B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810143432.6A CN108470146B (en) 2018-02-11 2018-02-11 Similar track identification method of classic track

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810143432.6A CN108470146B (en) 2018-02-11 2018-02-11 Similar track identification method of classic track

Publications (2)

Publication Number Publication Date
CN108470146A CN108470146A (en) 2018-08-31
CN108470146B true CN108470146B (en) 2022-07-08

Family

ID=63266545

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810143432.6A Active CN108470146B (en) 2018-02-11 2018-02-11 Similar track identification method of classic track

Country Status (1)

Country Link
CN (1) CN108470146B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108537258A (en) * 2018-03-26 2018-09-14 西南电子技术研究所（中国电子科技集团公司第十研究所） The similar flight path recognition methods of classical flight path
CN111506627B (en) * 2020-04-21 2023-05-30 成都路行通信息技术有限公司 Target behavior clustering method and system
CN117786428A (en) * 2024-02-27 2024-03-29 亿海蓝（北京）数据技术股份公司 Method, device and readable storage medium for determining ship track similarity

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105841708A (en) * 2016-03-16 2016-08-10 佛山科学技术学院 Vehicle navigation and positioning track matching method based on path tracing
CN105915224A (en) * 2016-04-07 2016-08-31 南京邮电大学 Parallelization track compression method based on Mapreduce
CN106778876A (en) * 2016-12-21 2017-05-31 广州杰赛科技股份有限公司 User classification method and system based on mobile subscriber track similitude
CN106960006A (en) * 2017-02-24 2017-07-18 河海大学 Measuring similarity system and its measure between a kind of different tracks
CN107271974A (en) * 2017-06-08 2017-10-20 中国人民解放军海军航空工程学院 It is a kind of based on the space-time error acquiring method for stablizing angle point
CN111967504A (en) * 2020-07-30 2020-11-20 西南电子技术研究所（中国电子科技集团公司第十研究所） Similar track judgment method with important point constraint

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170169297A1 (en) * 2015-12-09 2017-06-15 Xerox Corporation Computer-vision-based group identification

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105841708A (en) * 2016-03-16 2016-08-10 佛山科学技术学院 Vehicle navigation and positioning track matching method based on path tracing
CN105915224A (en) * 2016-04-07 2016-08-31 南京邮电大学 Parallelization track compression method based on Mapreduce
CN106778876A (en) * 2016-12-21 2017-05-31 广州杰赛科技股份有限公司 User classification method and system based on mobile subscriber track similitude
CN106960006A (en) * 2017-02-24 2017-07-18 河海大学 Measuring similarity system and its measure between a kind of different tracks
CN107271974A (en) * 2017-06-08 2017-10-20 中国人民解放军海军航空工程学院 It is a kind of based on the space-time error acquiring method for stablizing angle point
CN111967504A (en) * 2020-07-30 2020-11-20 西南电子技术研究所（中国电子科技集团公司第十研究所） Similar track judgment method with important point constraint

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
基于数据挖掘的四维航迹精密预测方法研究;马勇;《中国优秀博硕士学位论文全文数据库(硕士) 工程科技Ⅱ辑》;20170315(第03期);C031-1891 *
基于最长公共子序列的非同步相似轨迹判断;刘宇等;《电讯技术》;20171031;第57卷(第10期);1165-1170 *
经典轨迹的相似度量快速算法;王前东;《系统工程与电子技术》;20201031;第42卷(第10期);2189-2196 *
经典轨迹的鲁棒相似度量算法;王前东;《电子与信息学报》;20200831;第42卷(第8期);1999-2005 *

Also Published As

Publication number Publication date
CN108470146A (en) 2018-08-31

Similar Documents

Publication Publication Date Title
CN108470146B (en) Similar track identification method of classic track
Yang et al. Temporal dynamic appearance modeling for online multi-person tracking
Zhang et al. Comparison of similarity measures for trajectory clustering in outdoor surveillance scenes
Zhang et al. Output constraint transfer for kernelized correlation filter in tracking
Yang et al. Toward the repeatability and robustness of the local reference frame for 3D shape matching: An evaluation
US5644652A (en) System and method for automatic handwriting recognition with a writer-independent chirographic label alphabet
WO2016034008A1 (en) Target tracking method and device
Xu et al. Anchorface: An anchor-based facial landmark detector across large poses
CN107103326A (en) The collaboration conspicuousness detection method clustered based on super-pixel
Fan et al. Parallel tracking and verifying
Zhao et al. A KCF-based incremental target tracking method with constant update speed
CN105718918B (en) Aerial signature authentication method based on video
Li et al. Cascade Classifier Using Combination of Histograms of Oriented Gradients for Rapid Pedestrian Detection.
Bai et al. A hierarchical face recognition method based on local binary pattern
Okawa Online signature verification using a single-template strategy with mean templates and local stability-weighted dynamic time warping
CN108537258A (en) The similar flight path recognition methods of classical flight path
CN107389071B (en) Improved indoor KNN positioning method
Okawa Online signature verification using multi-distance measures and weighting with gradient boosting
Chuang et al. Human action recognition using star templates and delaunay triangulation
Shi et al. A real time vision-based hand gestures recognition system
CN114065851B (en) Maneuvering target track clustering method and system
Lei et al. Holoscopic 3D micro-gesture recognition based on fast preprocessing and deep learning techniques
Arbain et al. A review of triangle geometry features in object recognition
Chun et al. 3D star skeleton for fast human posture representation
CN116030059B (en) Target ID reauthentication matching method and system based on track

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