The Estimation of Traffic Flow Parameters based on Monitoring the

Speed Values using Computer Vision

V. D. Shepelev

, A. I. Vorobyev

, E. V. Shepeleva

, I. D. Alferova

, N. Golenyaev

G. Yakupova

and V. G. Mavrin

South Ural State University, 76 Lenin Prospekt, Chelyabinsk, Russia

Moscow Automobile and Road Construction State Technical University, 64, Leningradsky Prospect, Moscow, Russia

Kazan Federal University, 18 Kremlyovskaya str, Kazan, Russia

vadim_mmite@rambler.ru

Keywords: Monitoring, Neural Networks, Statistical Analysis, Traffic Capacity, Vehicle Speed, YOLOv3.

Abstract: Most of the previous works dealing with road traffic organization have been focused on optimizing the

setup of traffic signals, assuming that the traffic flow speed is fixed or adheres to a given distribution. In our

study, we focused on real-time determining the vehicle speed and assessing its influence on the vehicle

delay time. Vehicle detection and speed determination are based on real-time processing of video streams

by a convolutional neural network (YOLOv3). The developed system can identify and classify traffic flows

into eleven types, as well as track the motion path and speed of vehicles throughout the entire functional

area of a signal-controlled intersection. While analysing the data, we identified two important factors

corresponding to the presence of a queue of vehicles waiting for the green traffic light: 1. We identified the

nature and statistically significant measure of reducing the free vehicle movement speed, depending on the

size of the queue; 2. We determined the acceptable queue size, which does not affect the dynamics of

crossing the intersection by group vehicles moving from the previous intersection. The obtained data allows

us to optimize the operation of the adaptive traffic light control of intersections and to optimize the

synchronization of road network signals based on speed indications.

https://orcid.org/0000-0002-1143-2031

https://orcid.org/0000-0002-1890-6033

https://orcid.org/0000-0003-2080-3145

https://orcid.org/0000-0001-8484-8129

https://orcid.org/0000-0003-3657-3567

https://orcid.org/0000-0001-6822-3700

https://orcid.org/0000-0001-6681-5489

1 INTRODUCTION

Most of the previous studies on the optimization of

road traffic parameters (Wong et al., 2010; Wong

et al., 2011), including methods for the

synchronization of coordinated signals

(Skabardonis and Geroliminis, 2008; Liu et al.,

2011; Makarova et al., 2020) are focused on the

optimization of signal timings ignoring the speed

of traffic flows (Wu et al., 2015; Makarova et al.,

2017). The real-time movement speed depends on

several factors, including the condition and quality

of the road surface, the driver behaviour, the

vehicle performance characteristics, and road

conditions (Burkhardt et al., 2021; Tian et al.,

2021). Variable speed can be used on highways to

control the traffic flow to increase the traffic

capacity in open highway sections. In particular,

the procedure for detecting road accidents is

considered in (Allaby et al., 2007; Hadiuzzaman et

al., 2013; Škorput et al., 2010; Beymer et al.,

1997). In the studies focused on monitoring the

level of traffic flow emissions, the key input

variables are instantaneous measurements of the

752

Shepelev, V., Vorobyev, A., Shepeleva, E., Alferova, I., Golenyaev, N., Yakupova, G. and Mavrin, V.

The Estimation of Trafﬁc Flow Parameters based on Monitoring the Speed Values using Computer Vision.

DOI: 10.5220/0010539407520759

In Proceedings of the 7th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2021), pages 752-759

ISBN: 978-989-758-513-5

vehicle speed and acceleration (Ahn et al., 2002;

Pavlovic et al., 2021). Therefore, it is essential to

closely monitor the flow speed, which can facilitate

making effective management decisions, as well as

predicting the influence of vehicle emissions on the

ambient air quality (Agarwal and Mustafi, 2021).

There are methods for detecting anomalies and

interpreting road traffic analysis using the Global

Navigation Satellite System (GNSS). The method

is based on measuring the Euclidean distance

between the STM (Speed Transition Matrices)

centre of mass (COM) and the mean STM, which

represents normal driving conditions. Space and

time-related events are combined into so-called

traffic congestion propagation patterns. These

patterns provide a high-level description of traffic

congestion and its propagation in time and space

(Tisljaric et al., 2020; Wang et al., 2013). This

approach is characterized by a significant data

transmission delay and averaging, which does not

allow one to instantly receive and respond to any

deviations in traffic parameters.

The vehicle speed determines the time the

vehicle needs to cross the functional area of the

intersection. The authors of (Wang, 2007)

considered a general approach to the development

of universal means for assessing the state of road

traffic for highway sections based on stochastic

macroscopic traffic modelling and extended

Kalman filtering. Vakili et al. (2020), Czajewski

and Iwanowski (2010) present a speed calculation

method based on the use of geometric information

and the distance travelled by vehicles. The

algorithms are based on processing a video image

taken by a single camera on the road to extract the

license plate in the image.

In this study, we focused on developing a

method for traffic speed monitoring to determine

the delay time of the vehicle queue. The main

objectives of this paper are: 1) to highlight the

nature and statistically significant measure of

reducing the free vehicle movement speed in the

presence of a queue in front of a traffic light; 2) to

determine the size of the queue, which does not

affect the dynamics of crossing the intersection by

group vehicles.

Although speed is very important to ensure the

safety and efficiency of setting road traffic control

systems (Choi et al., 2013; Makarova et al., 2018),

and there are known advantages of integrating

speed into traffic signal synchronization programs

(Abu-Lebdeh, 2010; Daganzo and Pilachowski,

2011), there remain technical difficulties in a

reliable real-time determination of the driving

speed and interpretation of big data.

2 DEVELOPMENT OF A

METHOD OF DETERMINING

THE VEHICLE SPEED

Our approach is based on the use of street video

surveillance cameras with a viewing angle

providing visibility of the entire functional area of

the intersection and the adjacent roads (Online

broadcast). We used the architecture of the

YOLOv3 neural network, which consists of 106

layers and is a modification of the Darknet-53

neural network (Khazukov et al., 2020). Besides, it

includes 53 more layers with two N-dimensional

output layers providing for the detection at three

different scales. This modification contributes to

more accurate vehicle recognition and

classification. As input data, YOLOv3 accepts an

image represented as a three-dimensional tensor

h×w×3, where h, w is the height and length of the

input image. We used OpenCV open-source library

to work with machine vision algorithms and

process images and general-purpose numerical

algorithms. To track traffic flows, we used the Sort

library built on elementary data associations and

methods for assessing the state of objects. To

calculate the vehicle speed, we need to find the

distance travelled based on the change in the

latitude and longitude of the object location, using

the change in coordinates. To solve this problem,

we calculated the perspective transformation matrix

(1) by selecting reference points in the map and

comparing their corresponding points in the image

(Figures 1, 2) (Khazukov et al., 2020).

Figure 1: Reference points in the image.

The Estimation of Trafﬁc Flow Parameters based on Monitoring the Speed Values using Computer Vision

753

Figure 2: Reference points in the image.













′













(1)

where A is the perspective transformation matrix, x

are the pixel coordinates in the image; x´

, y´

are

the latitude and longitude of a point in the image.

We will calculate the distance between two

points based on the formula of inverse haversine

presented explicitly through the arcsine:

() ()

























−













−

⋅=

sincoscos

sin

arcsin2

λλ

ϕϕ

(

where D is the measured distance; φ

, λ

are the

latitude and longitude of the i-th point; r =6.371 km

is the earth radius.

We will find the average speed by the following

formula:

−

(3)

where t

, t

is the time of the beginning and the end

of the object movement at a distance.

The distance measurement accuracy is calculated

and verified based on a perspective transformation

determining the area of the used pixels transmitting

the road section (Figure 3).

The actual size of the marked distance is 88.5 m,

in the image this segment is transmitted as 91px.

One pixel in this section covers an area with a

length of 0.97 m. Taking this area as a square, we

Figure 3: The distance transmitted by pixels.

can estimate the maximum projection error of a

point within this area of 1.37 m. Thus, the error in

determining the speed does not exceed 1.7 km/h.

2.1 Statistical Significance of

Differences

In statistical analysis, it is generally accepted to

consider the revealed regularity to be statistically

significant when the empirical level of significance

is less than the generally accepted critical value of

0.05 (5%) (Byul, 2005; Tyurin and Makarov, 2016).

For the considered problem of estimating the

statistical significance of reducing the time spent on

crossing the intersection in the presence of a queue,

we should make sure that the calculated mean values

of time (Table 1), from the standpoint of statistics,

are not in the same confidence interval, i.e., they do

not represent the same numeric value.

We used the professional SPSS Statistical

Analysis Package for the calculations. According to

Table 1, the empirical significance levels of

deviations are much lower than the limiting ones,

which indicates the statistical significance of the

vehicle speed deviations in the presence of a queue

before the intersection.

The SPSS package additionally calculates paired

Pearson correlation coefficients between the same

variables (Table 2), which show a weak correlation

between them.

This additionally confirms the legitimacy of the

conclusion that the changes in the vehicle speeds are

statistically significant - due to the absence of

hidden regularities in empirical data, which could

distort the analysis results.

The recommended verification of the results of

the statistical significance of the differences in the

speed of the vehicles passing the intersection in the

interpretation of the nonparametric approach also

confirmed the high significance of its change. Table

3 presents the results of the calculations using the

Wilcoxon nonparametric signed-rank test.

iMLTrans 2021 - Special Session on Intelligent Mobility, Logistics and Transport

754

Table 1: Statistical significance of the deviations of the intersection crossing speeds.

Paired differences t Degrees

freedom

Significance

Mean Standard

deviation

Pair 1 Och1 & OchN 1.43704 1.33611 12.497 134 0.00

Pair 2 Och1 & NoOch1 2.02133 1.403367 16.732 134 0.00

Pair 3 Och1 & NoOchN 3.00489 1.37474 25.397 134 0.00

Pair 4 OchN & NoOch1 0.58430 1.11680 6.079 134 0.00

Pair 5 OchN & NoOchN 1.56785 0.96799 18.819 134 0.00

Pair 6 NoOch1 & NoOchN 0.98356 1.10166 10.373 134 0.00

Table 2: Correlation of paired samples.

N Correlation Significance

Pair 1 Och1 & OchN 135 0.230 0.007

Pair 2 Och1 & NoOch1 135 0.199 0.020

Pair 3 Och1 & NoOchN 135 0.067 0.438

Pair 4 OchN & NoOch1 135 0.284 0.001

Pair 5 OchN & NoOchN 135 0.286 0.001

Pair 6 NoOch1 & NoOchN 135 0.177 0.041

Table 3: Statistical significance by the Wilcoxon test.

Asymptotic

significance

(two-sided)

OchN - Och1 -8.584

0.00

NoOch1 - Och1 -9.456

0.00

NoOchN - Och1 -10.048

0.00

NoOch1 - OchN -5.515

0.00

NoOchN - OchN -9.644

0.00

NoOchN - NoOch1 -7.666

0.00

2.2 Analysis of the Vehicle Queue Size

This study is based on the video camera data on the

operation of 22 urban intersections and, similarly,

assumes the solution of the following tasks:

1. additional processing of the initial data to

make them homogeneous;

2. analysis of the mean values of the intervals

between the vehicles located in different

initial positions in the queue in front of a

traffic light when they enter the intersection;

3. determination of the queue size, when the last

vehicle in the queue passes the intersection as

if there is no delay in the queue.

2.2.1 Formation of a Homogeneous Sample

The previous clustering of the analysed intersections

showed a contrast in the empirical data for the two

intersections containing tram lines. Therefore, for

further analysis, we use about 30 observations for

each of the twenty intersections quite similar in the

way they are passed by vehicles. Eventually, for the

analysis in this task, we used 590 records of passing

the vehicle queue only on the green traffic light.

Similar to the first study, we removed the

observations for the transport categories other than

M1 (vehicles, which are used to carry passengers

and have no more than eight seats in addition to the

driver’s seat), M2 (vehicles, which are used to carry

passengers, have more than eight seats in addition to

the driver’s seat, the technically permissible

maximum mass of which does not exceed <5 t), and

N1 (small trucks with the technically permissible

maximum mass not exceeding <3.5 t.) from the

empirical data (Classification of vehicles according

to technical regulations, 2018). The size of the

vehicle queue considered in this study is limited by

the availability of a sufficient amount of the

empirical data – these are 13 vehicles. According to

the final results, it turned out to be sufficient to form

reliable conclusions.

The Estimation of Trafﬁc Flow Parameters based on Monitoring the Speed Values using Computer Vision

755

2.2.2 Mean Values of the Time Needed to

Pass the Intersections by the Vehicle

Queue

The summary calculated data on processing the

sample of observations of passing twenty

intersections by a vehicle queue are presented in

summary Table 4.

Notably, the table clearly shows the difference

between the intersections in the dynamics of

crossing by vehicles. This has been shown earlier

when the intersections were clustered into

homogeneous groups. However, in this task, this

difference does not affect the distortion of the

general trend but only emphasizes the generality of

the conclusions obtained for various intersections.

The processing results are graphically shown in

Figure 4, where the horizontal axis indicates the

position of the vehicle in the queue before the

intersection, and the vertical axis indicates the mean

time of the interval this vehicle needs to enter the

intersection.

Table 4: Mean time of the intervals the vehicles need to enter the intersection.

Intersections

Position of the vehicle in the queue

1 2 3 4 5 6 7 8 9 10 11 12 13

Prkr01 3.0 2.9 2.7 2.3 2.6 2.5 2.4 1.8 2.2 2.3 2.8 2.0 2.0

Prkr02 3.5 3.6 2.5 3.1 2.7 2.3 2.1 2.1 2.0 2.1 2.0 2.0 1.5

Prkr03 3.4 2.8 2.7 2.4 2.2 2.3 2.4 2.0 2.5 2.0 1.9 1.6 1.6

Prkr04 3.0 3.0 2.7 2.1 2.4 1.5

Prkr05 2.6 2.8 2.2 1.7 2.4 1.8 2.0 1.5 1.5 2.0

Prkr06 3.0 2.8 2.3 2.5 2.0 2.0 2.2 1.6 2.0

Prkr07 4.2 2.3 2.1 2.1 2.0 2.0 1.8 2.2 1.6

Prkr08 3.4 3.5 2.8 2.8 3.0 2.0 2.4 2.7 2.3 2.5

Prkr09 4.5 2.8 2.7 2.1 2.0 2.0

Prkr10 4.1 2.2 1.9 1.8 2.1 2.1 2.0 1.6 2.0 2.0 2.4 2.2 1.8

Prkr11 3.1 2.4 2.2 2.0 1.9 2.0 2.0 1.8 1.7 2.6 1.7 2.0 2.2

Prkr12 2.9 2.5 2.5 2.5 2.0 2.4 1.9 1.8 2.2 2.1 2.5 2.2 1.7

Prkr13 3.0 2.8 2.3 2.7 1.5 2.0

Prkr14 3.0 2.6 2.4 2.3 2.6 3.0 2.0 2.0 2.0

Prkr15 4.6 2.2 2.3 2.3 2.2 2.0

Prkr16 5.1 2.1 2.0 2.0 2.0 2.0

Prkr17 4.6 2.8 2.5 2.7 2.5 3.0 2.0

Prkr18 3.7 2.6 2.8 2.4 2.1 2.8 2,2 2.2 2.9 1.8 1.5 2.0 2.0

Prkr19 3.8 1.9 1.9 1.7 2.0 1.8 1.6 1.7 1.6 2.6 1.3 2.0

Prkr20 5.2 2.2 1.9 2.0 2.1 1.8 1.7 1.8 2.2 1.5 1.5

Mean: Sr1 Sr2 Sr3 Sr4 Sr5 Sr6 Sr7 Sr8 Sr9 Sr10 Sr11 Sr12 Sr13

3.72 2.69 2.41 2.31 2.24 2.19 2.07 1.96 2.08 2.16 1.97 2.01 1.86

Figure 4: The intervals a vehicle needs to enter the intersection, depending on its position in the queue.

iMLTrans 2021 - Special Session on Intelligent Mobility, Logistics and Transport

756

We can assume a priori that the dynamics of the

vehicles passing the intersection, starting from the

7th vehicle in the queue, becomes stable.

That is, a queue of six vehicles or more already

does not slow down the time of passing the

intersection by subsequent vehicles. The mean time

of the interval between vehicles entering the

intersection is two seconds.

However, these preliminary conclusions should

be confirmed in terms of their statistical

significance.

2.2.3 The Size of the Queue Stabilizing the

Interval between Vehicles

Taking into account the significant influence of the

human factor in fixing empirical data, as well as

their major gradation by an observer within one

second, a nonparametric approach used in similar

conditions will be more suitable for statistical

analysis. Moreover, the normal distribution of the

initial data is out of the question.

Notably, the samples are linked through the

observed intersections. Therefore, the statistical

analysis method most suitable in this study is the

Wilcoxon nonparametric signed-rank test for two

linked samples (SPSS statistical analysis package).

We will use this method to check the possible pairs

of differences of all the calculated mean values of

the Sr1-Sr13 intervals from the a priori assumed

SR0 value of two seconds.

The calculation results are presented in Table 5.

According to the calculations, the mean values,

starting from Sr6, fall into the confidence interval of

the a priori expected value of two seconds, i.e., they

Table 5: Statistical significance by the Wilcoxon test.

Asymptotic

significance

(two-sided)

SR0-Sr1 -3.922

SR0-Sr2 -3.885

SR0-Sr3 -3.696

SR0-Sr4 -2.940

0.3%

SR0-Sr5 -2.728

0.6%

SR0-Sr6 -1.790

7.4%

SR0-Sr7 -1.716

8.6%

SR0-Sr8 -1.171

24.2%

a. Positive ranks are used

b. Negative ranks are used

become statistically indistinguishable. This means

that in the queue before the intersection, vehicles,

starting from the 6th position in the queue, pass the

intersection with the time intervals corresponding to

the absence of a queue. This time interval

corresponds to the generally accepted estimates of

two seconds.

Notably, the mean values of the intervals

corresponding to the vehicles’ positions from 9 to 13

are not considered due to the decreasing amount of

the initial data and because their absolute values are

much closer to the a priori expected SR0 value than

for SR6.

3 CONCLUSIONS

In the course of the study, we identified two

important factors corresponding to the presence of a

vehicle queue before the intersection on the red

traffic light.

First, we revealed the nature and statistically

significant measure of reducing the free movement

vehicle speed in the presence of a queue in front of

the traffic light.

Second, we determined the size of the queue,

which does not affect the dynamics of passing the

intersection by the vehicles following the queue. We

also determined their mean interval of movement

equal to two seconds.

In addition to these two factors manifested and

studied in this paper, which correspond to the

presence of a vehicle queue, we can note other

interesting areas of research, such as a

heterogeneous structure of the category of vehicles

in the queue, their location in the queue, and several

other important situations. These areas are the

subject of our further research generally intended for

task-oriented vehicle flow management.

REFERENCES

Abu-Lebdeh, G., 2010. Exploring the potential benefits of

IntelliDrive-enabled dynamic speed control in

signalized networks. Proceedings of the 89th Annual

Meeting of the Transportation Research Board, # 10-

3031.

Agarwal, A.K., Mustafi, N.N., 2021. Real-world

automotive emissions: Monitoring methodologies, and

control measures. Renewable and Sustainable Energy

Reviews, 137(110624).

Ahn, K., Rakha, H., Trani, A., Van Aerde, M., 2002.

Estimating vehicle fuel consumption and emissions

based on instantaneous speed and acceleration levels.

The Estimation of Trafﬁc Flow Parameters based on Monitoring the Speed Values using Computer Vision

757

Journal of Transportation Engineering, 128 (2): 182-

190.

Allaby, P., Hellinga, B., Bullock, M., 2007. Variable

speed limits: Safety and operational impacts of a

candidate control strategy for freeway applications.

IEEE Transactions on Intelligent Transportation

Systems, 8 (4): 671-680.

Atev, S., Masoud, O., Janardan, R., Papanikolopoulos, N.,

2005. A collision prediction system for traffic

intersections. In Proceedings of the IEEE/RSJ

International Conference on Intelligent Robots and

Systems (IROS), 1545407: 169-174.

Beymer, D., McLauchlan, P., Coifman, B., Malik, J.,

1997. Real-time computer vision system for measuring

traffic parameters. In Proceedings of the IEEE

Computer Society Conference on Computer Vision

and Pattern Recognition, pp. 495-501.

Buch, N., Cracknell, M., Orwell, J., Velastin, S.A., 2009.

Vehicle localisation and classification in urban CCTV

Streams. In Proceedings of the 16th World Congress

on Intelligent Transport Systems and Services (ITS

2009)

Buch, N., Velastin, S.A., Orwell, J., 2011.A review of

computer vision techniques for the analysis of urban

traffic. IEEE Transactions on Intelligent

Transportation Systems, 12 (3) # 5734852: 920.

Buch, N., Yin, F., Orwell, J., Makris, D., Velastin, S.A.,

2009. Urban vehicle tracking using a combined 3D

model detector and classifier. Lecture Notes in

Computer Science, 5711 LNAI (PART 1): 169-176.

Buivol, P.A., Iakupova, G.A., Makarova, I.V.,

Mukhametdinov, E.M., 2020. Search and optimization

of factors to improve road safety. International Journal

of Engineering Research and Technology, 13

(11):3751-3756.

Burkhardt, M., Yu, H., Krstic, M., 2021. Stop-and-go

suppression in two-class congested traffic.

Automatica, 125(109381).

Byul, A., 2005. SPSS: the art of information processing.

Statistical data analysis and recovery of hidden

patterns. Moscow, DiaSoft.

Choi, J., Tay, R., Kim, S., 2013. Effects of changing

highway design speed. Journal of Advanced

Transportation, 47(2): 239-246.

Classification of vehicles according to technical

regulations (November 2018). Available at https://xn--

80aaf3axmme8h.xn--p1ai/registratsiya-i-uchet/klassifi

katsiya-ts

Czajewski, W., Iwanowski, M., 2010. Vision-based

vehicle speed measurement method. Lecture Notes in

Computer Science, 6374 LNCS (PART 1): 308-315.

Dailey, D.J., Cathey, F.W., Pumrin, S., 2000.An algorithm

to estimate mean traffic speed using uncalibrated

cameras. IEEE Transactions on Intelligent

Transportation Systems, 1 (2): 98-107.

Gunawan, A.A.S., Tanjung, D.A., Gunawan, F.E., 2019.

Detection of vehicle position and speed using camera

calibration and image projection methods. Procedia

Computer Science, 157: 255-265.

Hadiuzzaman, M., Qiu, T.Z., 2013. Cell transmission

model based variable speed limit control for freeways.

Canadian Journal of Civil Engineering, 40 (1): 46-56.

Khazukov, K., Shepelev, V., Karpeta, T., Shabiev, S.,

Slobodin, I., Charbadze, I., Alferova, I., 2020. Real-

time monitoring of traffic parameters. Journal of Big

Data, 7 (1), # 84.

Kim, H., 2019. Vehicle detection and speed estimation for

automated traffic surveillance systems at nighttime.

Tehnicki Vjesnik, 26 (1): 87-94.

Maduro, C., Batista, K., Batista, J., 2009. Estimating

vehicle velocity using image profiles on rectified

images. Lecture Notes in Computer Science, 5524

LNCS: 64-71.

Makarova, I., Pashkevich, A., Shubenkova, K., 2017.

Ensuring Sustainability of Public Transport System

through Rational Management. In Proceedings of the

16th International Scientific Conference Reliability

and Statistics in Transportation and Communication,

178: 137-146.

Makarova, I., Shubenkova, K., Mavrin, V., Buyvol, P.,

2018. Improving safety on the crosswalks with the

use of fuzzy logic, Transport Problems, 13 (1): 97-

109.

Makarova, I., Yakupova, G., Buyvol, P., Mukhametdinov,

E., Pashkevich, A., 2020. Association rules to identify

factors affecting risk and severity of road accidents. In

Proceedings of the 6th International Conference on

Vehicle Technology and Intelligent Transport Systems

(VEHITS), 614-621.

Online broadcast. Video surveillance. 2021. Availible at

https://cams.is74.ru/live

Pavlovic, J., Fontaras, G., Broekaert, S., Ciuffo, B.,

Ktistakis, M.A., Grigoratos, T., 2021. How accurately

can we measure vehicle fuel consumption in real

world operation? Transportation Research Part D:

Transport and Environment, 90(102666).

Škorput, P., Mandžuka, S., Jelušić, N., 2010. Real-time

detection of road traffic incidents. Promet - Traffic -

Traffico, 22 (4): 273-283.

Tian, J., Zhu, C., Chen, D., Jiang, R., Wang, G., Gao, Z.,

2021. Car following behavioural stochasticity analysis

and modeling: Perspective from wave travel time.

Transportation Research Part B: Methodological,

143:160-176.

Tisljaric, L., Majstorovic, Z., Erdelic, T., Caric, T., 2020.

Measure for traffic anomaly detection on the urban

roads using speed transition matrices. In Proceedings

of the 43rd International Convention on Information,

Communication and Electronic Technology (MIPRO),

9245327: 252-259.

Tyurin, Yu. N., Makarov, A. A., 2016. Data analysis on a

computer: textbook. Moscow, ICNMO.

Vakili, E., Shoaran, M., Sarmadi, M.R., 2020. Single–

camera vehicle speed measurement using the geometry

of the imaging system. Multimedia Tools and

Applications, 79 (27-28): 19307-19327.

Wang, Y., Papageorgiou, M., Messmer, A., 2007. Real-

time freeway traffic state estimation based on

iMLTrans 2021 - Special Session on Intelligent Mobility, Logistics and Transport

758

extended Kalman filter: A case study. Transportation

Science, 41 (2): 167-181.

Wang, Z., Lu, M., Yuan, X., Zhang, J., Wetering, H.V.D.,

2013. Visual traffic jam analysis based on trajectory

data. IEEE Transactions on Visualization and

Computer Graphics, 19 (12), 6634174: 2159-2168.

Wu, W., Li, P.K., Zhang, Y., 2015. Modelling and

simulation of vehicle speed guidance in connected

vehicle environment. International Journal of

Simulation Modelling, 14 (1): 145-157.

Young, C., Rice, J., 2006. Estimating velocity fields on a

freeway from low-resolution videos. IEEE

Transactions on Intelligent Transportation Systems, 7

(4): 463-469.

The Estimation of Trafﬁc Flow Parameters based on Monitoring the Speed Values using Computer Vision

759