You are on page 1of 5

2nd Int'l Conf.

on Electrical Engineering and Information & Communication Technology (ICEEICT) 2015


Jahangirnagar University, Dhaka-1342, Bangladesh, 21-23 May 2015

Real-time Area Based Traffic Density Estimation by


Image Processing for Traffic Signal Control System:
Bangladesh Perspective
2 3
Mohammad Shahab Uddin \ Ayon Kumar Das , Md. Abu Taleb
Department of Electrical and Electronic Engineering
Chittagong University of Engineering and Technology
Chittagong, Bangladesh
mohammadshahabuddin.eee@gmail.com1, ayonkd@gmail.com2, abu.taleb048@gmail.com3

Abstract-Traffic congestion is a daily occurrence in most algorithms are applied to control traffic jam in [15-16].In [17-
urban areas of Bangladesh now a day. In the last 10 years the 21], Neural networks have been proposed to control traffic
scenario has worsen due to rapid increase of vehicles and lights to solve traffic congestion. Reinforcement learning[22-
insufficient roads to accommodate them. This paper describes a
23] is also proposed for traffic signal control system. "Prewit"
method of real time area based traffic density estimation using
edge detection method is used to control traffic in [24]. We
image processing for intelligent traffic control system. Area
occupied by the edges of vehicles will be considered to estimate
have proposed an area based technique for the detection of
vehicles density. Calculating the areas of different live roads, the traffic density using image processing for intelligent traffic
system will automatically estimate the traffic density of each road light control system. This technique can better estimate the
which will help to determine the duration of each traffic light. An traffic density using the area occupied by the edges of
intelligent traffic signal control system with the proposed traffic vehicles. Area based traffic density estimation will be more
density estimation technique will be far better than the effective for controlling traffic lights than the conventional
conventional timer based system of Bangladesh. The main methods for the countries like Bangladesh where different
contribution of this research lies in the development of a new
types of vehicles running on the same road.
technique that detects traffic density according to the area of the
edges of vehicles for controlling traffic congestion. Specialized
algorithm, morphology and images captured with cameras will
III. METHODOLOGY

be used for the detection of traffic density for the intelligent Our method for detecting traffic congestion involves the
traffic control system.
following steps:
Keywords-Image processing, edge detection, traffic congestion A) Empty Road Analysis:
detection, intelligent traffic control system
1) Acquisition of empty road images.
2) Image cropping as reference.
I. INTRODUCTION
3) ROB to ORAY conversion
Traffic congestion is one of the major problems of urban
4) Removal of unwanted object
life. This problem is increasing day by day because of the
increasing number of vehicles with limited infrastructural 5) Edge detection using canny method
development. Under this circumstance, the conventional 6) Dilation of images
traffic light systems which are timer based are not able to 7) Area measurement of reference images.
control traffic congestion. If a lane has more traffic congestion
than the others, the existing system fails to control traffic. To
solve this problem, a real time traffic control system is needed
which will control the traffic light according to traffic density.
In this paper we have proposed a real time area based traffic
density estimation method which will help an intelligent
traffic control system to control traffic light according to
traffic density.

II. LITERATURE REVIEWS

A lot of techniques are proposed to detect traffic density to


control traffic signal. Image processing is one of the most
popular techniques. Many methods[I-4] are used to detect
traffic density using image processing. Fuzzy logic[5-14] has
been used to control traffic signals. Moreover, evolutionary Fig. 1. Image acquisition & processing

978-1-4673-6676-2/15/$31.00 ©2015 IEEE


B) Road with Traffic Analysis: posItIOn of our camera is fixed so we can easily remove
1) Acquisition of road images with traffic. unwanted objects like trees, footpaths, shops, buildings by
2) Image cropping. cropping the road picture. These objects create obstacle in our
measurement and make some unwanted errors. So we used the
3) ROB to ORAY conversion
cropping image tool to eliminate the errors. Fig. 4 shows the
4) Removal of unwanted object cropped original images of empty roads and fig. 5 shows the
5) Edge detection using canny method cropped original image of live roads with vehicles.
6) Dilation of images
7) Area measurement of live images.

C) Decision Making:
Compare the areas and find the difference. There would be
surely a difference between empty road and road with
vehicles. The program decides which road should be given
more priority based on cars density.

IV. SYSTEM IMPLEMENTATION

A) Image Acquisition: Fig. 4. Cropped empty road


Images of empty roads and roads with vehicles were taken
with cameras .Fig. 2 shows the original image of empty roads
and fig. 3 shows the original image of roads with vehicles.

Fig. 5. Cropped live road

Fig. 2. Image of empty roads C) RGB to GRAY Conversion:


Color (ROB) image converted to gray format to detect the
edges correctly. Fig. 6 shows the gray images of original
empty roads and fig. 7 shows the gray original image of live
roads with vehicles.

Fig. 3. Image of roads with vehicles

B) Image Cropping:
Cropping of original image helps us to reduce our Fig. 6. Cropped empty road (gray)

processing time as well as to analyze our region of interest


more easily. It increases the accuracy of calculation. Since the
Fig. 7. Cropped live road with traflic (gray)
Fig. 9. Edge detection of live road

D) Removal of Unwanted Objects:


F) Image Dilation:
There are many unwanted objects like trees, buildings etc.
on both sides of the roads. These objects will decrease the After applying Canny edge detection method there are gaps
accuracy of the measurement of traffic density. For this between the edge segments of the images. These broken edges
reason, we have made these objects seem to be disappeared do not dilate the outline of the vehicles clearly. We have used
using fill operation. dilation operation to fill in the gaps between edge segments
preserving the details of the vehicles. Fig. 10 and fig. 11
shows the dilated images of empty and live roads.
E) Edge Detection using Canny Method:
We have used "Canny" edge detection method [25] for
detecting edge of the images. This detection method is widely
used in image processing for edge detection. It has some
advantages over the other methods, which are as follows [25-
27]:
• It maximizes the signal-to-noise ratio.
• It minimizes multiple responses to a single edge.
• It ensures that the edge points are well localized.

Canny edge detection method has the following steps [25-


27]:
• Noise is removed with the Gaussian filter. Fig. 10. Dilated mask of empty road
• Gradient's module as well as direction is computed.
• Non-maximum suppression ensures that local
maxima are marked as edges.
• Adaptive hysteresis thresholding is used for edge
linking.
Fig. 8 and fig. 9 shows the output after applying canny
edge detection method.

Fig. 11. Dilated mask of live road

G) Comparison ofAreas:
After applying dilation operation, we have compared the
areas of the live images with the areas of the reference images.
The images of empty roads are taken as reference images.
Fig. 8. Edge detection of empty road
According to our method traffic signal will be controlled by
measuring areas of edges of the vehicles. Traffic congestion
will be estimated according to the area. Road having greater
area of the edges of vehicles will be considered as more traffic
40000 ,-----
congested than others. Fig. 12 represents the area of empty
road, live road and the area occupied by the edges of vehicles. 35000 +-------==--
Fig. 13 shows the flow chart of whole area based traffic 30000
density estimation method where AI, A2, A3, A4 indicates the
25000
area of edges of vehicles for road I, road 2, road 3, and road 4 • FULL
20000
respectively. Fig. 14 shows the relationship between the • EMPTY

vehicles number and the area of the edges of vehicles. All the 15000 .AREA

steps of this method are implemented by using the MATLAB 10000


software package.
5000

ROAD! ROAD2 ROAD3 ROAD4

Fig. 12. Bar representation

Yes No

No

Yes

Fig. 13. Flow chart of traflic congestion estimation process


25
Transportation Engineers 65th Annual Meeting, pp. 154-158,August
-1 '9 04 , � )= 1995.
20 [10] Qinghui Lin, B.W. Kwan, and LJ. Tung, "Traffic signal control using
u
lil fuzzy logic," in Proc. IEEE Conf Syst, Man, Cybem., vol. 2,pp. 1644-
15
'-
0 1649,October 1997.
� ( :3.1>5 0., =8) [11] T M. Sayers, MG.H. Bell, T Mieden, and F. Busch, "Improving the
.D
10
§ traffic responsiveness of signal controllers using fuzzy logic," in Proc.
Z -S-�:i9,$ IEE Colloq. Urban Congestion Management, pp. 6/1-6/4,November
5 --
-2;; S- , 1995.
0 [12] Jarkko Nittymaki, "Installation and experience of field testing a fuzzy
signal controller," EUL 1. Operat Res., voL 131, pp. 273-281,June 2001.
o 5000 10000 15000 20000
[13] Jee-Hyong Lee, and Hyung Lee-Kwang, "Distributed and cooperative
Area of edges (pixel) fuzzy controllers for traffic intersections group," in IEEETrans. Syst,
Man,Cybem, pt C, voL 29, no. 2, pp. 263-271,May 1999.
[14] W. Wen," A dynamic and automatic traffic light control expert system
Fig. 14. Graphical representation for solving the road congestion problem", Expert Systems with
Applications, VoL 34(4),pp. 2370-2381,May 2008.
[15] Shaw C Chang, Ming W. Tsai, and Gi W. Huang ,"A GA based
V. CONCLUSION intelligent traffic signal scheduling model," in IEEE Symposium on
Computational Intelligence in Scheduling, pp. 93-97,2007.
In this paper, we have proposed a traffic density estimation
[16] DONG Chaojun, L1U Zhiyong, and L1U Xian kun, "Chaos-particle
technique using image processing based on the area occupied swarm optimization algorithm and its application to urban traffic
by the edges of vehicles. The proposed method can easily control," International Journal of Computer Science and Network
estimate traffic density according to experiment results. This Security, vol. 6(IB),pp. 97-101,2006.
[17] Dipti Srinivasan, Min Chee Choy, and Ruey Long Cheu,"Neural
area based traffic density estimation method can be easily
networks for real-time traffic signal control," IEEE Transactions on
implemented in an intelligent traffic control system in a dense Intelligent Transportation Systems, vol. 7(3),pp. 261-272,September
populated country. Our proposed method will play an 2006.
important role in estimating traffic congestion to control [18] Jiuyi Hua, and Ardeshir Faghri, "Development of neural signal control
system: toward intelligent traffic signal control,"in Transportation
traffic signals but it still requires development to attain higher
Research Record 1497, Trans. Res. Board, Washington, DC, pp. 53-61,
accuracy. July 1995.
[19] R. K. Saraf,"Adaptive traffic control using neural networks," PhD.
Dissertation, Dept Civil Environ. Eng., Vanderbilt Univ., Nashville,
TN, 1994.
[20] D. C Chin, l C Spall, and R. H. Smith, "Evaluation of system-wide
REFERENCES
traffic signal control using stochastic optimization and neural net-
[I] Rita Cucchiara, Massimo Piccardi, and Paola Mello, "Image analysis works," in Proc. 1999 American Control Conf, voL 3,pp. 2188-
and rule-based reasoning for a traffic monitoring system," IEEE 2194,June 1999.
Transaction on Intelligent Transportation System, voL 1,no. 2,June [21] Dipti Srinivasan, Min Chee Choy, and Ruey Long Cheu, "Neural
2000. network for real-time traffic signal control", IEEE Transactions on
[2] M Fathy, and MY Siyal, "Real-time image processing approach to Intelligent Transportation Systems, vol. 7,no. 3,pp. 261-272,September
measure traffic queue parameters,"IEE Proc.-Vis. Image Signal 2006.
Process,voL 142,no. 5, pp. 297-303,October 1995. [22] Michael L Littman, and Csaba Szepesvari,"A generalized reinforcement
[3] Yoichiro Iwasaki, "An image processing system to measure vehicular learning model: convergence and applications," in 13th International
queues and an adaptive traffic signal control by using the information of Conference Machine Learning,pp. 310-318, 1996.
the queues," in IEEE conference on Intelligent Transportation
[23] Zhenjiang Li, Feng He, Qingming Yao, and Fei-Vue Wang, " Signal
System,pp. 195-200,November 1997.
controller design for agent-based traffic control system," in IEEE
[4] 1.M. Blosseville, C Krafft,F. Lenoir,V. Motyka, and S. Beucher,
International Conference on Networking, Sensing and Control, pp. 199-
'TITAN: a traffic measurement system using image processing
204, April 2007.
techniques,"in Second International Conference on Road Traffic
[24] Madhavi Arora, and V. K. Banga, "Real Time Traffic Light Control
Monitoring,pp. 84-88,February 1989.
System", 2nd International Conference on Electrical, Electronics and
[5] C P. Pappis and Ebrahim H. Mamdani, "A fuzzy logic controller for a
Civil Engineering, pp. 172-176, April 2012.
traffic junction," IEEE Transactions on Systems, Man, and Cybernetics
,voL7, pp.707-717,October 1977. [25] John Canny, "A computational approach to edge detection",IEEE Trans.
Van Li and Xiaoping FAN," Design of signal controllers for urban Pattern Anal. Machine IntelL, vol. PAMI-8, no. 6, pp. 679-
[6]
intersections based on fuzzy logic and weightings," in 6th IEEE 697,November 1986.
Conference on Intelligent Transportation Systems, voL 1, pp. 867- [26] Jamil AM Saif,AIi Abdo Mohammed AI-Kubati,Abdultawad Saif
871,October 2003. Hazaa, and Mohammed AI-Moraish,"Image Segmentation Using Edge
[7] G. Nakamiti and F. Gomide, "Fuzzy sets in distributed traffic control," Detection and Thresholding",in proc. Of 13th international arab
in Proc. 5th IEEE Int Conf Fuzzy Systems, voL 3, pp. 1617- conference on Information Technology, December 2012.
1623,September 1996. [27] Tinku Acharya, and Ajoy K. Ray, Image Processing: Principles and
[8] Stephen Chiu and Sujeet Chand, "Self-organizing traffic control via Application, Hoboken: John Wiley and Sons,2005.
fuzzy logic," in Proc. 32nd IEEE Conf Decision Control, pp. 1987-
1902,December 1993.
[9] A Tzes, W. R. McShane, and S. Kim, "Expert fuzzy logic for traffic
signal control for transportation networks," in Proc. Institute of

You might also like