Volume 7, Issue 3, March – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Density Based Traffic Control using Image

Processing Methods: A Review
Narendra Mohan#1, Ghanshyam Mishra*2
M.Tech Scholar#1, Assistant Professor *2
Buddha Institute of Technology, Gorakhpur

Abstract:- Congestion on the roads can result in fatigue, length of time it remains green in the opposite direction is
increased fuel consumption, energy and time waste, typically determined by simple scheduling and calculated
increased depreciation and fuel costs, air pollution, when the intersection is designed.
accidents, and verbal and physical altercation among the
commuters. Cost, untrustworthy technology, and the Current real-time traffic signal control technology is
inability of the contemporary system to incorporate expensive and frequently inaccurate. Thus, the emphasis is
elements of the new technology are the few of the placed on this critical point, and every effort is made to be
challenges which has to be dealt with. Advanced image fair.
processing algorithms, such as segmentation and edge
detection, can be used to detect vehicle presence in It has the potential to significantly reduce daily traffic
different lanes and calculate traffic density. Thus, India's by smoothing traffic flows and providing real-time priority to
traffic management system can be improved and the busiest lane. The system can be configured to handle any
enhanced using low-cost solutions. This paper presents a unexpected traffic flow.
review of various techniques for automated density-based In this paper a review on various types of prominent
traffic management. image forgery and tampering techniques have been discussed
Keywords:- Traffic Control, Density based traffic along with the work done till date in the past decade by
management, Image processing, edge detection. researchers in the area of image forgery detection and image
forensics has been discussed.
I. INTRODUCTION
II. RELATED WORK
Congestion on the road creates havoc and makes it
difficult for commuters to travel. Many major cities are A. Kanungo et al. [1] reduced the image matrix by
grappling with the problem. Congestion in traffic can have a estimating vehicle and empty space in the path, calculating
significant impact on a country's economy. This problem is congestion in all red directions, and signalling the green light
caused by a variety of factors, the most common of which are to the most congested path using a MATLAB simulator.
poor traffic management, cars changing lanes, unplanned Congestion on a previously green path is reduced to zero,
stoppages, and so on. Congestion on the roads can cause whereas congestion on two red lights in opposite directions
fatigue, increased fuel consumption, increased depreciation remains unchanged. The number of vehicles that can pass
and fuel costs, considerable wastage of energy as well as time. through each congestion per second is also factored into the
There are a number of problem areas in the domain of smart green signal schedule. This design improved by 35% when
traffic control system such as high implementation cost, compared to the standard situation, in which the green light
unreliable technology and rigidity of the existing system in signal is given the same amount of time in both directions.
place. The issue with this traditional system is that traffic M.M. Hasan et al. [2] proposed an image-processing-
light scheduling is done regardless of instantaneous traffic based method for determining road traffic congestion, as well
density detection. As a result, even when there is no vehicle as a model for controlling traffic signals using data from
passing in one direction, the schedule of the passage for that road-camera images. Instead of counting vehicles, they
direction is considered, and time is wasted. Existing traffic calculate traffic density, which is the total area of the road
control and management systems are obsolete and incapable occupied by vehicles in terms of total pixels in a video frame.
of dealing with the various situations that arise as a result of They have set two output parameters based on traffic density
different levels of traffic congestion throughout the day. for each road, variable traffic cycle and weighted time, and
Another issue in this area is the scheduling and prevention of control traffic lights sequentially.
traffic accidents at intersections. The length of time a traffic
light stays green in one direction and how long it stays green Image processing, according to D.Sitaram et al. [3], can
in the opposite direction is typically determined by simple be used in situations where the surveillance camera is too
scheduling and calculated when the intersection is designed. slow to send the live video in a fraction of a second, or where
the server of the website that displays the live feed from these
Existing traffic control and management systems are surveillance cameras is too slow. They access websites
obsolete and incapable of dealing with the various situations hosted by the Public Traffic Information System, which
that arise throughout the day as a result of varying levels of display real-time traffic images captured by surveillance
traffic congestion. Another issue in this area is the scheduling cameras installed in various locations. The traffic density is
and prevention of intersection traffic accidents. The length of calculated by assigning a rating on a 10-point scale ranging
time a traffic light remains green in one direction and the from 1 to 10.

IJISRT22MAR423 www.ijisrt.com 279

Volume 7, Issue 3, March – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
N.R.Islam et al. [4] proposed implementing a smart image (TSI) processing. The TSI is a technique for
traffic control system based on traffic density measurement combining multiple video images into a single image. As a
using a real-time video processing technique. To determine result, the TSI can save memory resources when compared to
the most effective method, video sequences from a camera traditional methods. A camera can be used for traffic-state
are analysed using object detection and counting methods. In estimation by integrating with TSI generating and processing
order to control the traffic signal brilliantly, the computed modules. Variables such as space-mean-speed, flow, and
vehicle density is compared to other parts of the traffic. RFID density can also be estimated. Several experiments' empirical
sensors are used in the system to ensure that law enforcement results are presented to show that TSI processing is a viable
is carried out. As a result, any vehicle or car breaking traffic low-cost approach to traffic state estimation.
laws can be easily identified.
P.Faldu et al. [9] proposed a real-time traffic
T. Tahmid and E. Hossain [5] proposed a traffic control information collection and monitoring system to address the
system that measures real-time vehicle density using canny issue of traffic road management. The proposed model,
edge detection and digital image processing. This which is a function of real-time traffic density, sequentially
commanding traffic control system outperforms existing controls the clearance time of each lane. The approach is
systems in terms of response time, vehicle management, hybrid in nature, combining sensor networks and camera
automation, reliability, and overall efficiency. Furthermore, technology. The combination of image processing and IOT
the entire technique is depicted with proper schematics, from yields greater accuracy and efficiency than the traditional
image acquisition to edge detection to green signal allotment, approach. RFID allows for the prioritisation of emergency
and the final results are validated by hardware vehicles, as well as the reduction of traffic congestion. Other
implementation, using four sample images of different traffic technologies are also used, such as stolen vehicle
conditions. tracking/detection. A simple user interface improves the
system by making it easier to manage in the control room.
K. V. Najiya and M. Archana [6] proposed a novel
framework for estimating traffic flow parameters from aerial In the absence of an emergency vehicle, S. S. P. Moka
videos using improved vehicle detection. To improve et al. [10] proposed and implemented a smart traffic
contrast, the frames are first subjected to Adaptive Gamma surveillance system in which access will be granted to the
Correction. The proposed method employs the Kanade- lane with the highest traffic density. When an emergency
Lucas-Tomasi (KLT) tracker, the Support Vector Machine vehicle is detected in any lane, the lane in which the
(SVM), and connected graphs. The KLT tracker is used to emergency vehicle is detected is given priority. The presence
analyse motion based on key points of interest. Vehicles are of an emergency vehicle is detected using signal processing
distinguished from other moving objects using the SVM techniques, and the density of traffic is computed using
classification. Two of the seven features extracted from the digital image processing techniques. The entire proposed
frames are the contourlet transform coefficients and the Gray system is depicted with appropriate schematics, from image
Level Co-occurrence matrix (GLCM). Finally, the number of and sound acquisition to canny edge detection and noise
vehicles on the road, the average speed, and the bidirectional removal using a Least Mean Square (LMS) filter, and finally
flow densities are computed. green signal allocation to lanes based on the output results.

According to Okaishi et al. [7], vision-based III. TRAFFIC CONTROL USING IMAGE PROCESSING
surveillance systems are widely used in traffic data analysis
due to their ease of installation and accuracy. A real-time Image processing is any type of signal processing in
image processing system for detecting and classifying which an image, such as a photograph or video frame, is used
vehicles at intersections is proposed in this paper. This data as input and the output is either an image or a set of image-
can be used to estimate traffic density at intersections and related characteristics or parameters. The vast majority of
adjust traffic light timing for the next light cycle. The image-processing techniques involve treating the image as a
background subtraction technique is used in the detection two-dimensional signal and applying standard signal-
operation, and the approximated median filter is used to processing techniques to it. A road image can be represented
extract and update the background, after which the vehicles as digital data, or binary data, but it must be processed before
are tracked in the detection area. Then, to classify vehicles, a it can be used to extract useful information. This is required
convolutional neural network will be used. because images captured in their natural environment are raw
and unformatted. Figure 1 depicts a basic framework used in
Ua-areemitr[8] proposed a low-cost image processing the majority of image-based traffic density measurement
system for estimating road traffic state based on time-spatial techniques.

IJISRT22MAR423 www.ijisrt.com 280

Volume 7, Issue 3, March – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

Image Image Pre-

Acquisition processing Edge Detection

Comparison in all
Control Traffic lanes Density Calculation

Fig. 1: General Framework for Traffic Control System using Image Processing

Image Acquisition: Cameras are used in all of the Image Preprocessing: Preprocessing is done to obtain a
methods discussed in this paper for image acquisition. To clear image. Because the images are extracted from real-time
capture the view of the traffic scene on the road, a camera is video frames, they may be distorted, blurred, or dark, as in
installed at any tall structures, such as polls, at traffic foggy or rainy weather. Similarly, images taken at night can
intersections. The captured video from the camera is then be too dark, while images taken in bright sunlight can be too
used to extract image frames. These frames are then analysed bright (like in afternoon). As a result, depending on the user's
and preprocessed in order to detect and count the number of goal, different preprocessing methods are used on images to
vehicles present, also known as vehicle density. improve image quality. Figure 2 depicts the original acquired
image (a) and the preprocessed image (b)[5].

Fig. 2: (a): Original Image (b) Preprocessed Image

Edge Detection: Techniques for detecting and available, including canny edge detection, sobel edge
distinguishing vehicles from the background are used. There detection, and prewitt edge detection. Figure 3 depicts the
are several thresholding and edge detection algorithms image following edge detection[5].

Fig. 3: Image post Edge Detection

IJISRT22MAR423 www.ijisrt.com 281

Volume 7, Issue 3, March – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Density Calculation: This is done to determine the density Detection," 2018 Second International Conference on
of vehicles in each lane. Several techniques, such as histogram- Inventive Communication and Computational
based thresholding and morphological operations such as Technologies (ICICCT), 2018, pp. 662-668, doi:
dilation, erosion, and filling, have been presented. Edge 10.1109/ICICCT.2018.8473204.
detection methods were also used to estimate the density within [7.] W. AL OKAISHI, A. ZAARANE, I. SLIMANI, I.
the lanes. ATOUF and M. BENRABH, "A Traffic Surveillance
System in Real-Time to Detect and Classify Vehicles by
Based on the preceding steps, the density of vehicles in Using Convolutional Neural Network," 2019 International
each lane can be estimated, and then decisions can be made for Conference on Systems of Collaboration Big Data,
effective traffic management and control. Internet of Things & Security (SysCoBIoTS), 2019, pp. 1-
5, doi: 10.1109/SysCoBIoTS48768.2019.9028037.
IV. CONCLUSIONS [8.] E. Ua-areemitr, A. Sumalee and W. H. K. Lam, "Low-
Image processing is used to automatically estimate and Cost Road Traffic State Estimation System Using Time-
control traffic density, which is critical for traffic management Spatial Image Processing," in IEEE Intelligent
in megacities. Congestion on the roads is becoming a major Transportation Systems Magazine, vol. 11, no. 3, pp. 69-
problem. Traffic congestion can be caused by a variety of 79, Fall 2019, doi: 10.1109/MITS.2019.2919634.
factors, including ineffective transportation management, a lack [9.] P. Faldu, N. Doshi and R. Patel, "Real Time Adaptive
of traffic information, and so on. Traditional methods for Traffic Control System : A Hybrid Approach," 2019 IEEE
estimating traffic density, such as radars, loop sensors, 4th International Conference on Computer and
ultrasonic waves, and so on, have some limitations, such as Communication Systems (ICCCS), 2019, pp. 697-701,
high cost, sensitivity to external environmental conditions, doi: 10.1109/CCOMS.2019.8821749.
lighting conditions, and so on. The use of computer vision [10.] S. S. P. Moka, S. M. Pilla and S. Radhika, "Real Time
technology can aid in the resolution of the problem. Advanced Density Based Traffic Surveillance System Integrated
image processing algorithms, such as segmentation and edge with Acoustic Based Emergency Vehicle Detection,"
detection, can be used to detect vehicle presence in different 2020 4th International Conference on Computer,
lanes and calculate traffic density. Thus, India's traffic Communication and Signal Processing (ICCCSP), 2020,
management system can be improved and enhanced using low- pp. 1-7, doi: 10.1109/ICCCSP49186.2020.9315209.
cost solutions. This paper contains a thorough discussion and
review of various concepts, methods, and work done by
previous researchers in the field of density-based traffic control.

REFERENCES

[1.] Kanungo, A. Sharma, and C. Singla, “Smart traffic lights

switching and traffic density calculation using video
processing,” 2014 Recent Adv. Eng. Comput. Sci., pp. 1–
6, 2014
[2.] M. M. Hasan, G. Saha, A. Hoque and Md. Badruddoja
Majumder, "Smart traffic control system with application
of image processing techniques," 2014 International
Conference on Informatics, Electronics & Vision (ICIEV),
2014, pp. 1-4, doi: 10.1109/ICIEV.2014.6850751.
[3.] D. Sitaram, N. Padmanabha, Supriya S and Shibani S,
"Still image processing techniques for intelligent traffic
monitoring," 2015 Third International Conference on
Image Information Processing (ICIIP), 2015, pp. 252-255,
doi: 10.1109/ICIIP.2015.7414775.
[4.] M. R. Islam, N. I. Shahid, D. T. ul Karim, A. A. Mamun
and M. K. Rhaman, "An efficient algorithm for detecting
traffic congestion and a framework for smart traffic
control system," 2016 18th International Conference on
Advanced Communication Technology (ICACT), 2016,
pp. 802-807, doi: 10.1109/ICACT.2016.7423566.
[5.] T. Tahmid and E. Hossain, "Density based smart traffic
control system using canny edge detection algorithm for
congregating traffic information," 2017 3rd International
Conference on Electrical Information and Communication
Technology (EICT), 2017, pp. 1-5, doi:
10.1109/EICT.2017.8275131.
[6.] K. V. Najiya and M. Archana, "UAV Video Processing
for Traffic Surveillence with Enhanced Vehicle

IJISRT22MAR423 www.ijisrt.com 282