with Intervehicle Communications and Sparse Video Surveillance Cameras Yang Wang Zhiwei Lv Wuji Ch eng He ngchang Liu University of Science and Technology of China ID: 782714
Download The PPT/PDF document "Tracking Urban Private Vehicles" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Tracking Urban Private Vehicles
with Inter-vehicle Communications and Sparse Video Surveillance Cameras
Yang Wang,
Zhiwei
Lv
,
Wuji
Ch
eng, He
ngchang
LiuUniversity of Science and Technology of China
Slide2SECON 18
System ModelVehicle Tracking System
Numerical Evaluation
201
8
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Basic Setting
Conclusion
Basic Setting
System Model
Numerical Evaluation
Vehicle Tracking System
Conclusion
SECON 18
System Model
Basic Setting
Slide3SECON 18
System ModelVehicle Tracking System
Numerical Evaluation
201
8
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Basic Setting
Conclusion
Basic Setting
System Model
Numerical Evaluation
Vehicle Tracking System
Conclusion
Slide4201
8
O
nline telematics systems
Traffic jams
& Traffic accidents
Sparse distribution of road video surveillance cameras
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide5201
8
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide6201
8
M
AIN
CONTRIBUTION
We
targeting the tracking problem of all urban private vehicles with V2V communications and sparse video surveillance cameras , meanwhile taking the influences of the time parameter into consideration while abstracting the patterns
of private vehicles.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide7201
8
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Conclusion
Basic Setting
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
System Model
Basic Setting
Vehicle Tracking System
Numerical Evaluation
Conclusion
Slide8201
8
B
ASIC
DATA
MODELING
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Conclusion
Basic Setting
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide9201
8
P
ROBLEM
DESCRIPTION
Accuracy Ratio AR of all the tracks of all urban private vehicles:
（
2.1
）
（
2.2
）
Accuracy Ratio AR:
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Conclusion
Basic Setting
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide10201
8
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Vehicle Tracking System
System Model
Numerical Evaluation
Conclusion
Basic Setting
Slide11201
8
T
HE
ARCHITECTUR
OF
OUR
SYSTEM
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Fig 3
.
1:System architectur
Slide12201
8
A
NALYSIS
OF
V
EHICLE
M
OVING
P
ATTERNS
T
he distance between the two corresponding trajectories：
Fig 3
.
2:
Heat map of traffic volumes and direction of road segments
（
3.1
）
（
3.2
）
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide13201
8
T
RAVEL
T
IME
-
COSTS
ANALYSIS
·
road segment travel time-cost follow
an
exponential distribution
.
·
Kolmogorov-Smirnov test(K-S test) check travel time of a road segment
·
The probability density function of the road segment travel time is shown
right
：
Fig 3
.
3:
K-S test
（
3.3
）
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide14201
8
V2V COMMUNICATION OPTIMIZATION PROBLEM
The number of seed cars is very small. Therefore, we need more large-scale urban private cars to acquire more vehicle trajectory information through dedicated short range communications(DSRC).
Fig 3
.
4:
Basic pattern diagram
D
ata collection
Data analysis
Provide services
Data analysis
platform
Vehicles with online telematics systems
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide15201
8
D
ATA
TRANSMSSION
WITH
THE
VEHICULAR
NETWORK
and then the following score function can be used:
（
3.4
）
（
3.5
）
（
3.6
）
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide16201
8
D
ATA
TRANSMSSION
WITH
THE
VEHICULAR
NETWORK
（
3.7
）
T
he redundancy degree
can be
represented as
:
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide17201
8
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Numerical Evaluation
Basic Setting
System Model
Vehicle Tracking System
Conclusion
Slide18201
8
E
VALUATION
S
ETUP
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide19201
8
P
ERFORMANCE
OF
THE
C
ANOPY
AND
K-
MEANS
COMBINED CLUSTERING ALGORITHM
As shown in this figure, there exist obvious differences among the trajectories of the corresponding 4 vehicle clusters during
different time periods on different types
of days.
Fig 4
.
1:
Heat map of vehicle trajectories of 4 vehicle clusters during different time slots in different day
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide20201
8
P
ERFORMANCE
OF
THE
C
ANOPY
AND
K-
MEANS
COMBINED CLUSTERING ALGORITHM
Fig 4
.
2:
Average encounter time between two vehicles
T
he
encounter time
between two vehicles in a same cluster is always obviously longer than the encounter time between two vehicles in different clusters
;
C
lustering preprocessing can
enhance the performances
of our algorithm significantly, and the reason is that it can help to estimate the time that a vehicle spends to meet a Wi-Fi hotspot in the future efficiently
.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide21201
8
C
OMPARISON
A
LGORITHM
C
omparison
Algorithm
E
xplain
FIFO
First
I
n First Out
Algorithm
PRLB
P
acket
R
esidual
L
ifetime-based
Algorithm
PhotoNet
Algorithm
B
ased on
D
istance
C
onstraint
PVT
Private Vehicle Tracking with Inter-vehicle Communications and
Sparse Video Surveillance Cameras
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide22201
8
I
MPACTS
OF
TIME
PERIODS
Fig 4.3:
Impacts of time periods on workdays
Fig
4.4:
Impacts of time periods on rest days
T
he performance of our solution is always better than the other three approaches
on both workdays and rest days.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide23201
8
I
MPACTS
OF
ROAD
TRAFIC
STATUSES
Fig 4.5:
Impacts of road traffic statuses
O
ur solution outperforms the other three approaches in both rush hours and normal hours
；
Because the number of vehicles participate in transmissions is quite large,the performances of all approaches are quite stable
.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide24201
8
I
MPACTS
OF
PERCENTAGES
OF
VEHICLES
PARTICIPATE
IN TRANSMISSIOPNS
Fig 4
.
6:
Impacts of percentages of vehicles participate in transmissions
O
ur approach outperforms the other three alternative solutions significantly
；
When only 50% of vehicles participate in transmissions, our solution can still ensure that the average AR value of tracking trajectories of all active vehicles is greater than 65%
.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide25201
8
P
ACKET
UPLOAD
DIFERENCES
BETWEEN
DIFFERENT
W
I
-FI SPOTS
Fig 4
.
7:
Average numbers of received packets of different Wi-Fi spots
I
n Wi-Fi spots 3 and 4, the numbers of collected packets of the four solutions are almost the same, but in spots 1 and 2, the differences between the numbers of received packets of different strategies are quite obvious
；
C
ompared with the other three alternative solutions, our solution’s collected packet number
is the least, the reason is that our solution can greatly benefit from calculating the expected time for vehicles from one cluster to meet a fixed Wi-Fi spot in the future.
SECON 18
System Model
Vehicle Tracking System
Numerical Evaluation
Basic Setting
Conclusion
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide26201
8
Vehicle Tracking System
Numerical Evaluation
Conclusion
SECON 18
System Model
Basic Setting
Vehicle Tracking System
SECON 18
System Model
Basic Setting
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Conclusion
Basic Setting
System Model
Vehicle Tracking System
Numerical Evaluation
Slide27201
8
C
ONCLUSION
Offers a new perspective to track
u
rban
p
rivate
v
ehicles with sparse video surveillance cameras and Inter-vehicle Communications
.
Model the travel time-cost with
exponential distribution
, then use
exponential distribution
to approximate the time-cost of an urban trip.
Tracking
u
rban
p
rivate
v
ehicles problem of this paper
is
converted to V2V
communication optimization problem using
a novel method
.
Use real-world vehicle and video surveillance information datasets to demonstrate the effectiveness of our algorithm
.
Numerical Evaluation
Conclusion
Vehicle Tracking System
SECON 18
System Model
Basic Setting
SECON 18
Tracking Urban Private Vehicles with Inter-vehicle Communications and SparseVideo Surveillance Cameras
Slide28201
8
.0
6
THANKS
Q&A
(