CN114021629A - Vehicle track motion mode extraction method based on mean value dynamic time warping - Google Patents

Vehicle track motion mode extraction method based on mean value dynamic time warping Download PDF

Info

Publication number
CN114021629A
CN114021629A CN202111248749.4A CN202111248749A CN114021629A CN 114021629 A CN114021629 A CN 114021629A CN 202111248749 A CN202111248749 A CN 202111248749A CN 114021629 A CN114021629 A CN 114021629A
Authority
CN
China
Prior art keywords
track
cluster
distance
motion mode
vehicle
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
CN202111248749.4A
Other languages
Chinese (zh)
Other versions
CN114021629B (en
Inventor
鲍虎军
华炜
杨非
叶娇娇
杜承垚
高飞
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.)
Zhejiang University of Technology ZJUT
Zhejiang Lab
Original Assignee
Zhejiang University of Technology ZJUT
Zhejiang Lab
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 Zhejiang University of Technology ZJUT, Zhejiang Lab filed Critical Zhejiang University of Technology ZJUT
Priority to CN202111248749.4A priority Critical patent/CN114021629B/en
Publication of CN114021629A publication Critical patent/CN114021629A/en
Application granted granted Critical
Publication of CN114021629B publication Critical patent/CN114021629B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • 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
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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

Abstract

The invention discloses a vehicle track motion mode extraction method based on mean value dynamic time warping. Initializing and resampling an original track to obtain a resampled track set; calculating the distance between tracks based on mean value dynamic time warping according to the resampling tracks; a track-to-track distance measurement method based on mean dynamic time warping, and clustering by using a clustering algorithm to obtain a cluster set; and finally, respectively extracting a characteristic motion mode for each cluster to obtain a characteristic motion mode set. According to the invention, by adopting a limited technology, the distance between similar tracks can be accurately measured, the distance between long-sequence tracks can be more accurately measured, the method is suitable for the vehicle behavior analysis of long-sequence tracks, and a characteristic motion mode can be extracted from the vehicle track data set; each extracted characteristic motion mode track can intuitively reflect the motion mode of the vehicle, and the motion mode is easier to be endowed with semantic endowing and visualized displaying.

Description

Vehicle track motion mode extraction method based on mean value dynamic time warping
Technical Field
The invention relates to a vehicle track motion mode extraction method based on mean dynamic time warping, in particular to a method for extracting a track motion mode when a vehicle passes through an intersection by using an inter-track distance measurement method of mean dynamic time warping.
Background
With the rapid development of Chinese economy, traffic is developed more and more, and the traffic problems appearing together with the development of the economy are endless. The problems of vehicle congestion, vehicle violation and discipline, traffic security and the like also emerge before people; therefore, the way of performing data cleaning, mining and classification on the vehicle track is an important means for analyzing the vehicle behavior at present.
The vehicle trajectory acquisition modes are numerous, and include an earlier and more mature inter-frame difference method, an optical flow field-based method, a background difference-based method, and a more emerging target detection and tracking method based on deep learning. However, no matter what method is adopted to identify and acquire the vehicle track and the implication in the track is wanted to be mined, the track behavior analysis is needed. The most common analysis of trajectory behavior is through trajectory classifiers, including DTW-based classification methods, LSTM methods based on recurrent neural networks, HMM-based methods, methods based on sequential probability gaussian models, and the like. Although the common methods can realize the track classification through the classification model obtained by training, the information between long sequences is still easy to lose, and the distance between tracks cannot be accurately measured, so that the distance between similar track groups is excessively large in pairs, and the vehicle tracks of each category are not easy to visually display; therefore, in order to overcome the problems of the traditional classification model, the invention provides the inter-track measurement method of mean value dynamic time warping on the basis of dynamic time warping, and extracts the track which can represent each category most from the vehicle track samples by combining the unsupervised clustering algorithm.
Disclosure of Invention
In view of the above problems in the prior art, the present invention is directed to a vehicle trajectory motion pattern extraction method based on mean dynamic time warping. The method measures the distance between the tracks according to the mean value dynamic time warping method, can accurately measure the distance between the similar tracks, enables the distance between the similar track groups to be close to each other, is more accurate in measuring the distance between the long sequence tracks, is particularly suitable for analyzing the behavior of the vehicles at the crossroads based on the long sequence tracks, and extracts the motion mode of the vehicle tracks.
The vehicle track motion mode extraction method based on mean value dynamic time warping is characterized by comprising the following steps of:
step 1: sample preparation: let the original vehicle trajectory sample data set be T ═ Ti1, 2., m }, where T is |, i ═ 1,2i＝{(xij,yij)|j＝1,2,...,miDenotes the ith original vehicle trajectory, (x)ij,yij) Represents the coordinates of the center point of the vehicle at the jth position in the ith original vehicle track, miRepresenting the coordinate quantity of the midpoint points of the ith original vehicle track, and m represents the sample quantity of the vehicle tracks; for all TiResampling to F at the same interval deltai＝{fis|s＝1,2,…,eiWhere δ denotes the resampled trace point interval, FiA resampled trace representing the ith trace,
Figure BDA0003321701310000021
s-th resampling track coordinate, resampling coordinate, representing ith track
Figure BDA0003321701310000022
P-th track representing i-th tracksA coordinate, psThe formula is shown as formula (1), wherein eiRepresenting the number of resampling coordinates of the ith track;
Figure BDA0003321701310000023
step 2: input resampling trajectory FiCalculating the distance between tracks based on mean value dynamic time warping, specifically: to measure the trajectory TiAnd the track TjDistance between, Ti∈T，TjE.g. T, j is not equal to i, firstly, constructing a distance matrix
Figure BDA0003321701310000024
And quantity matrix
Figure BDA0003321701310000025
And is
Figure BDA0003321701310000026
Wherein a is 1,2, …, ei，b＝1,2,…,ejAnd is and
Figure BDA0003321701310000027
representing points of track fiaAnd fjbThe distance between the two or more of the two or more,
Figure BDA0003321701310000028
representing a track TiFrom fi1To fiaSub-sequence and track T ofjFrom fj1To fjbThe number of times the point of DTW calculation is included in the subsequence of (a); to DijAnd OijCarry out initialization, i.e. for all
Figure BDA0003321701310000029
Performing a zeroing operation and recording according to a 'and b' of formula (2)
Figure BDA00033217013100000210
Subscript of medium minimum; then, D is calculated according to the formula (3) in a traversal modeijAll of the elements in
Figure BDA00033217013100000211
The result of (1); then, according to the formula (4), the O is calculated in a traversal wayijAll of the elements in
Figure BDA00033217013100000212
The result of (1); finally, according to the mean value dynamic time warping metric function output dis (T) of the formula (5)i,Tj) As a track TiAnd TjThe inter-track distance of (a);
Figure BDA0003321701310000031
Figure BDA0003321701310000032
Figure BDA0003321701310000033
Figure BDA0003321701310000034
and step 3: a track distance measurement method based on mean dynamic time warping clusters by using a DBSCAN clustering algorithm to obtain a cluster set C, which specifically comprises the following steps:
step 3.1: according to the inter-track distance measurement in the step 2, clustering is carried out by using a DBSCAN algorithm to obtain a clustering result set label ═ { label ═ labeli1,2, …, m, where labeliRepresenting a track TiClustering results of (2) and labeliE { -1,1,2, …, K }, where K represents the number of clusters, if labeliWhen it is-1, it represents the track TiIs an outlier trajectory; if labeliK denotes the track TiBelongs to the kth cluster, wherein K is 1,2, …, K;
step 3.2: constructing a clustering cluster set C ═ { C ═ C according to the clustering result set label and the sample set Tk1,2, …, K }, wherein the clusters
Figure BDA0003321701310000037
A set of tracks representing the k-th cluster,
Figure BDA0003321701310000038
indicating the ith vehicle track TiAnd the track belongs to cluster Ck；
And 4, step 4: cluster C ═ CkRespectively extracting characteristic motion mode CT from each cluster in the motion imagekAnd obtaining a characteristic motion mode set CT ═ { CT ═k1,2, …, K, where CTkThe characteristic motion pattern of the kth cluster is represented, specifically: for arbitrary cluster CkTraverse all traces T belonging to the clusteri kCalculating the average distance between the cluster and all other tracks of the cluster, and finally selecting the track with the minimum average distance as the characteristic motion mode of the cluster
Figure BDA0003321701310000036
The specific formula is shown as formula (6), wherein nkRepresents a cluster CkThe number of tracks in (1); and extracting characteristic motion patterns of all the clusters in the C to obtain a characteristic motion pattern set CT.
Figure BDA0003321701310000035
By adopting the detection, compared with the prior art, the invention has the following beneficial effects:
the invention discloses a vehicle track motion mode extraction method based on mean dynamic time warping, which measures the distance between vehicle tracks by using a mean dynamic time warping method, accurately measures the distance between similar tracks, enables the distances between similar track groups to be close in pairs, measures the distance between long sequence tracks more accurately, simultaneously utilizes a clustering method to complete clustering while rejecting outlier tracks, enables the track density in each cluster to be higher, can represent a track motion mode, extracts characteristic motion modes in the clusters, achieves the advantages of visual display of the motion mode and easiness in semantic endowment, and is particularly suitable for the behavior analysis of vehicles at intersections based on tracks.
Drawings
FIG. 1 is a sample set of traces illustrating an embodiment of the present invention;
FIG. 2 is a schematic view of the clusters of FIG. 1 after the tracks are classified;
fig. 3 is a schematic diagram of a trace motion pattern extracted from a trace sample data set according to an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and examples of the specification. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The invention discloses a vehicle track motion mode extraction method based on mean dynamic time warping, which measures the distance between tracks according to a mean dynamic time warping method, can accurately measure the distance between similar tracks, enables the distances between similar track groups to be close in pairs, measures the distance between long sequence tracks more accurately, is particularly suitable for analyzing the behavior of vehicles at a crossroad based on the long sequence tracks, and extracts the vehicle track motion mode; the vehicle track sample data visualization prepared by the embodiment of the invention is shown in fig. 1, and the result of extracting the characteristic motion mode by the method of the invention is shown in fig. 3, in the embodiment, the vehicle track motion mode extraction method based on mean value dynamic time warping specifically comprises the following steps:
step 1: sample preparation makes the original vehicle track sample data set as T ═ Ti1, 2., m }, where T is |, i ═ 1,2i＝{(xij,yij)|j＝1,2,...,miDenotes the ith original vehicle trajectory, (x)ij,yij) Represents the coordinates of the center point of the vehicle at the jth position in the ith original vehicle track, miRepresenting the coordinate number of the midpoint points of the ith original vehicle track, wherein m represents the sample number of the vehicle track, and in the embodiment of the invention, m is 492; for all TiResampling to F at the same interval deltai＝{fis|s＝1,2,…,eiWhere δ denotes the resampled trace point interval, FiA resampled trace representing the ith trace,
Figure BDA0003321701310000041
s-th resampling track coordinate, resampling coordinate, representing ith track
Figure BDA0003321701310000042
P-th track representing i-th tracksA coordinate, psThe formula is shown as formula (1),
Figure BDA0003321701310000051
wherein eiRepresenting the number of resampling coordinates of the ith track;
step 2: input resampling trajectory FiCalculating the distance between tracks based on mean value dynamic time warping, specifically: to measure the trajectory TiAnd the track TjDistance between, TiAnd TjAre all elements of T, i.e. Ti∈T，TjE.g. T, j is not equal to i, firstly, constructing a distance matrix
Figure BDA0003321701310000052
And quantity matrix
Figure BDA0003321701310000053
And is
Figure BDA0003321701310000054
Figure BDA0003321701310000055
Wherein a is 1,2, …, ei，b＝1,2,…,ejAnd is and
Figure BDA0003321701310000056
representing points of track fiaAnd fjbThe distance between the two or more of the two or more,
Figure BDA0003321701310000057
representing a track TiFrom fi1To fiaSub-sequence and track T ofjFrom fj1To fjbThe number of times the point of DTW calculation is included in the subsequence of (a); to DijAnd OijCarry out initialization, i.e. for all
Figure BDA0003321701310000058
Figure BDA0003321701310000059
Performing a zeroing operation and recording according to a 'and b' of formula (2)
Figure BDA00033217013100000510
Subscript of medium minimum; then, D is calculated according to the formula (3) in a traversal modeijAll of the elements in
Figure BDA00033217013100000511
The result of (1); then, according to the formula (4), the O is calculated in a traversal wayijAll of the elements in
Figure BDA00033217013100000512
The result of (1); finally, according to the mean value dynamic time warping metric function output dis (T) of the formula (5)i,Tj) As a track TiAnd TjThe inter-track distance of (a);
Figure BDA00033217013100000513
Figure BDA00033217013100000514
Figure BDA00033217013100000515
Figure BDA00033217013100000516
and step 3: a track distance measurement method based on mean dynamic time warping clusters by using a DBSCAN clustering algorithm to obtain a cluster set C, which specifically comprises the following steps:
step 3.1: according to the inter-track distance measurement in the step 2, clustering is carried out by using a DBSCAN algorithm to obtain a clustering result set label ═ { label ═ labeli1,2, …, m, where labeliRepresenting a track TiClustering results of (2) and labeliE { -1,1,2, …, K }, where K denotes the number of clusters, and in the example of the present invention, the number of clusters obtained by clustering K is 9, and the trajectory of each cluster after trajectory classification is shown in fig. 2, if labeliWhen it is-1, it represents the track TiIs an outlier trajectory; if labeliK denotes the track TiBelongs to the kth cluster, wherein K is 1,2, …, K;
step 3.2: and constructing a clustering cluster set C ═ { C ═ C according to the clustering result set label and the sample data set Tk1,2, …, K }, wherein the clusters
Figure BDA0003321701310000063
A set of tracks representing the k-th cluster,
Figure BDA0003321701310000064
indicating the ith vehicle track TiAnd the track belongs to cluster Ck；
And 4, step 4: set C ═ C for cluster clusteringkRespectively extracting characteristic motion mode CT from each cluster in the motion imagekAnd obtaining a characteristic motion mode set CT ═ { CT ═k1,2, …, K, where CTkThe characteristic motion pattern of the kth cluster is represented, specifically: for arbitrary cluster CkTraverse all vehicle trajectories belonging to the cluster
Figure BDA0003321701310000065
Calculating the average distance between the cluster and all other tracks of the cluster, and finally selecting the track with the minimum average distance as the characteristic motion mode of the cluster
Figure BDA0003321701310000061
The specific formula is shown as formula (6), wherein nkRepresents a cluster CkThe number of tracks in (1);
Figure BDA0003321701310000062
and extracting characteristic motion patterns from all clusters in the cluster set C to obtain a characteristic motion pattern set CT.
After the method is implemented, the distance between similar tracks can be accurately measured, the distance between long-sequence tracks can be more accurately measured, the method is suitable for analyzing the vehicle behavior of the long-sequence tracks, and a characteristic motion mode can be extracted from the vehicle track data set; each extracted characteristic motion mode track can intuitively reflect the motion mode of the vehicle, and the motion mode is easier to be endowed with semantic endowment and visualized display; in this example, the trajectory characteristic motion pattern extracted by using the method of the present invention is shown in fig. 3, which undoubtedly provides the vehicle motion pattern with functions of easy semantic assignment and visual display of the motion pattern.

Claims (2)

1. A vehicle track motion mode extraction method based on mean value dynamic time warping is characterized by comprising the following steps:
step 1: sample preparation makes the original vehicle track sample data set as T ═ Ti1, 2., m }, where T is |, i ═ 1,2i＝{(xij,yij)|j＝1,2,...,miDenotes the ith original vehicle trajectory, (x)ij,yij) Represents the coordinates of the center point of the vehicle at the jth position in the ith original vehicle track, miRepresenting the coordinate quantity of the midpoint points of the ith original vehicle track, and m represents the sample quantity of the vehicle tracks; for all TiResampling to F at the same interval deltai＝{fis|s＝1,2,…,eiWhere δ denotes the resampled trace point interval, FiA resampled trace representing the ith trace,
Figure FDA0003321701300000011
s-th resampling track coordinate, resampling coordinate, representing ith track
Figure FDA0003321701300000012
P-th track representing i-th tracksA coordinate, psThe formula (2) is shown as formula (1);
Figure FDA0003321701300000013
wherein eiRepresenting the number of resampling coordinates of the ith track;
step 2: input resampling trajectory FiCalculating the distance between tracks based on mean value dynamic time warping, specifically: for measuring original vehicle track TiAnd the track TjDistance between, Ti∈T，TjE.g. T, j is not equal to i, firstly, constructing a distance matrix
Figure FDA0003321701300000014
And quantity matrix
Figure FDA0003321701300000015
And is
Figure FDA0003321701300000016
Wherein a is 1,2, …, ei，b＝1,2,…,ejAnd is and
Figure FDA0003321701300000017
representing points of track fiaAnd fjbThe distance between the two or more of the two or more,
Figure FDA0003321701300000018
representing a track TiFrom fi1To fiaSub-sequence and track T ofjFrom fj1To fjbThe number of times the point of DTW calculation is included in the subsequence of (a); to pairDijAnd OijCarry out initialization, i.e. for all
Figure FDA0003321701300000019
Figure FDA00033217013000000110
Performing a zeroing operation and recording according to a 'and b' of formula (2)
Figure FDA00033217013000000111
Subscript of medium minimum; then, D is calculated according to the formula (3) in a traversal modeijAll of the elements in
Figure FDA00033217013000000112
The result of (1); then, according to the formula (4), the O is calculated in a traversal wayijAll of the elements in
Figure FDA00033217013000000113
The result of (1); finally, according to the mean value dynamic time warping metric function output dis (T) of the formula (5)i,Tj) As a track TiAnd TjThe inter-track distance of (a);
Figure FDA0003321701300000021
Figure FDA0003321701300000022
Figure FDA0003321701300000023
Figure FDA0003321701300000024
and step 3: based on a track-to-track distance measurement method of mean dynamic time warping, clustering is carried out by using a DBSCAN clustering algorithm to obtain a cluster set C;
and 4, step 4: cluster C ═ CkRespectively extracting characteristic motion mode CT from each cluster in the motion imagekAnd obtaining a characteristic motion mode set CT ═ { CT ═k1,2, …, K, where CTkThe characteristic motion pattern of the kth cluster is represented, specifically: for arbitrary cluster CkTraverse all traces T belonging to the clusteri kCalculating the average distance between the cluster and all other tracks of the cluster, and finally selecting the track with the minimum average distance as the characteristic motion mode of the cluster
Figure FDA0003321701300000025
The calculation formula is shown as formula (6), and after the characteristic motion patterns are extracted from all the clusters in C, a characteristic motion pattern set CT is obtained,
Figure FDA0003321701300000026
wherein n iskRepresents a cluster CkNumber of tracks in (1).
2. The method for extracting vehicle trajectory motion mode based on mean value dynamic time warping as claimed in claim 1, wherein the step 3 is a trajectory distance measurement method based on mean value dynamic time warping, and a DBSCAN clustering algorithm is used for clustering, and obtaining a cluster set comprises:
step 3.1: according to the mean value dynamic time warping distance measurement in the step 2, clustering is carried out by using a DBSCAN algorithm, and a clustering result set label ═ { label is obtainedi1,2, …, m, where labeliRepresenting a track TiClustering results of (2) and labeliE { -1,1,2, …, K }, where K represents the number of clusters, if labeliWhen it is-1, it represents the track TiIs an outlier trajectory; if labeliK denotes the track TiBelongs to the kth cluster, wherein K is 1,2, …, K;
step 3.2: constructing a clustering cluster set C ═ { C ═ C according to the clustering result set label and the sample set Tk1,2, …, K }, wherein the clusters
Figure FDA0003321701300000031
Set of tracks representing the kth cluster, Ti kIndicating the ith vehicle track TiAnd the track belongs to cluster Ck。
CN202111248749.4A 2021-10-26 2021-10-26 Vehicle track motion mode extraction method based on mean dynamic time warping Active CN114021629B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111248749.4A CN114021629B (en) 2021-10-26 2021-10-26 Vehicle track motion mode extraction method based on mean dynamic time warping

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111248749.4A CN114021629B (en) 2021-10-26 2021-10-26 Vehicle track motion mode extraction method based on mean dynamic time warping

Publications (2)

Publication Number Publication Date
CN114021629A true CN114021629A (en) 2022-02-08
CN114021629B CN114021629B (en) 2024-04-05

Family

ID=80057676

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111248749.4A Active CN114021629B (en) 2021-10-26 2021-10-26 Vehicle track motion mode extraction method based on mean dynamic time warping

Country Status (1)

Country Link
CN (1) CN114021629B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110399686A (en) * 2019-07-30 2019-11-01 中国民航大学 A kind of unrelated aircraft flight profiles clustering method of parameter based on silhouette coefficient
CN110570658A (en) * 2019-10-23 2019-12-13 江苏智通交通科技有限公司 Method for identifying and analyzing abnormal vehicle track at intersection based on hierarchical clustering
CN111986235A (en) * 2020-08-25 2020-11-24 浙江工业大学 Method for extracting vehicle track characteristic motion mode
CN111986226A (en) * 2020-08-25 2020-11-24 浙江工业大学 Target track abnormity detection method
WO2021077760A1 (en) * 2019-10-23 2021-04-29 江苏智通交通科技有限公司 Abnormal driving early warning method on basis of reasonable driving range of vehicle at intersection

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110399686A (en) * 2019-07-30 2019-11-01 中国民航大学 A kind of unrelated aircraft flight profiles clustering method of parameter based on silhouette coefficient
CN110570658A (en) * 2019-10-23 2019-12-13 江苏智通交通科技有限公司 Method for identifying and analyzing abnormal vehicle track at intersection based on hierarchical clustering
WO2021077760A1 (en) * 2019-10-23 2021-04-29 江苏智通交通科技有限公司 Abnormal driving early warning method on basis of reasonable driving range of vehicle at intersection
WO2021077761A1 (en) * 2019-10-23 2021-04-29 江苏智通交通科技有限公司 Intersection abnormal vehicle trajectory identification and analysis method based on hierarchical clustering
CN111986235A (en) * 2020-08-25 2020-11-24 浙江工业大学 Method for extracting vehicle track characteristic motion mode
CN111986226A (en) * 2020-08-25 2020-11-24 浙江工业大学 Target track abnormity detection method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张勇;张建伟;韩云祥;: "一种改进的航迹聚类方法", 现代计算机, no. 18, 25 June 2020 (2020-06-25) *
汪祖云;张道航;刘文平;董婉青;侯彩霞;陈荔;: "基于轨迹段核密度的旅游车辆轨迹聚类算法", 测控技术, no. 09, 18 September 2020 (2020-09-18) *

Also Published As

Publication number Publication date
CN114021629B (en) 2024-04-05

Similar Documents

Publication Publication Date Title
Kim et al. Gaussian process regression flow for analysis of motion trajectories
CN102609735B (en) Method and apparatus for assessing standard fulfillment of character writing
CN110070074B (en) Method for constructing pedestrian detection model
CN106875424B (en) A kind of urban environment driving vehicle Activity recognition method based on machine vision
CN103605362B (en) Based on motor pattern study and the method for detecting abnormality of track of vehicle multiple features
CN103390278B (en) A kind of video unusual checking system
CN110660083B (en) Multi-target tracking method combined with video scene feature perception
CN109410586A (en) A kind of Traffic State Detection Method based on multivariate data fusion
CN108171112A (en) Vehicle identification and tracking based on convolutional neural networks
CN102663429B (en) Method for motion pattern classification and action recognition of moving target
CN110120218A (en) Expressway oversize vehicle recognition methods based on GMM-HMM
CN105654139B (en) A kind of real-time online multi-object tracking method using time dynamic apparent model
CN107103326A (en) The collaboration conspicuousness detection method clustered based on super-pixel
CN107748873A (en) A kind of multimodal method for tracking target for merging background information
CN103854027A (en) Crowd behavior identification method
CN111986235A (en) Method for extracting vehicle track characteristic motion mode
CN104318263A (en) Real-time high-precision people stream counting method
CN109341703A (en) A kind of complete period uses the vision SLAM algorithm of CNNs feature detection
CN106570490B (en) A kind of pedestrian's method for real time tracking based on quick clustering
CN113378891A (en) Urban area relation visual analysis method based on track distribution representation
CN103336967B (en) A kind of hand motion trail detection and device
CN104050453A (en) Evaluation method for handwritten Chinese character handwriting
CN113256690B (en) Pedestrian multi-target tracking method based on video monitoring
CN114882069A (en) Taxi track abnormity detection method based on LSTM network and attention mechanism
CN109284705A (en) A kind of method for detecting abnormality of Traffic Surveillance Video

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