intelligent transportation system by controlling traffic...

Intelligent transportation system by controlling traffic using video

processing in Mat Lab

Vijay .J

Associate professor Department of ECE,

JeppiaarMaamallan Engineering College, Sriperumbudur.

Aloiciousarockiakoncy A

Department of ECE, JeppiaarMaamallan Engineering College,

Sriperumbudur.

Abstract: Computer vision techniques are used for

analysis of traffic surveillance videos which is

gaining more importance. This analysis of videos

can be useful for public safety and for traffic

management. In recent time, there has been an

increased scope for analysis of traffic activity

automatically. Computer based surveillance

algorithms and systems are used to extract

information from the videos which is also called as

Video analytics. The process of identifying

instances of real world objects is known as object

detection. It detects the number of vehicles on

each road and depending on the vehicles load on

each road, this system assigns optimized amount

of waiting time (red signal light) and running time

(green signal light). This system is a fully

automated system that can replace the

conventional pre-determined fixed-time based

traffic system with a dynamically managed traffic

system.

Keywords: Object detection, video analysis, bounding box, holes filling, KNN classifier.

1. Introduction

Every individual in homes have their own transportation vehicle for variety of purposes. In the current context of smart city, specifically in the industrial and market zones, the traffic scenario is very congested most of the time particularly at the peak time of business hours. Due to increasing growth of population and vehicles in smart and metropolitan cities people face lot of problem at the major traffic points of the business towns. To keep away from such severe issues many radiant urban communities are right now implementing smart traffic control frameworks that work on the standards of traffic automation with prevention of traffic issues. The fundamental concept lies in collection of traffic congestion information quickly and passing the alternate strategy to vehicles as well as passengers,

Chandraleka V.S


Sriperumbudur.

Gracelin D Jenibha


Sriperumbudur.

with on-line traffic information system and effectively applying it to specific traffic stream.By using the object detection method the vehicles are detected and they are counted for the density in different lanes. Prioritising the density in descending order the lanes are turned green for a specific period of time, following them is the another lane which stands next in the priority. Thus the traffic can be controlled automatically using the modern digital images or frames.

2. Literature survey

There are many literature works available on intelligent transportation system (ITA). The Intelligent Transportation System (ITS) provides services related to different modes of transport and traffic management systems with an integration of traffic control centres. Video-Based investigation for traffic surveillance has been a vital part of ITS. The traffic surveillance in urban environment have become more challenging compared to the highways due to various factors like camera placement, cluttered background, pose variation, object occlusion and illumination changes.[1]

The background subtraction technique is used

to find foreground objects. To detect the moving vehicles, thresholding, hole filling and adaptive morphology are applied. The vehicles were detected and counted with their virtual detection zone. But the drawback is that the virtual detection zone must be large. For shadow detection and shadow removal masking techniques is implemented.[2]

Later the vehicles are identified using

MSER feature detection, where correspondences between image elements from two images with different viewpoints. Feature matching compares one feature of image to another image to detect vehicle. In future generalization of vehicle must be taken into account. [3]

International Journal of Applied Engineering Research ISSN 0973-4562 Volume 14, Number 6, 2019 (Special Issue) © Research India Publications. http://www.ripublication.com

Page 307 of 310

mailto:[email protected]




To overcome this canny edge detection

method is applied where the edges are detected and the image is converted into binary image. Here the white pixels are calculated and compared with reference image. The matching percentage is calculated in which the matching percentage is directly proportional to the time delay.[4] Later background subtraction is done by Gaussian Mixture model(GMM) where the vehicles are detected by blob analysis. Vehicles are counted by incrementing counter by bounding box for vehicle. The disadvantage is that this method fails to removes the shadow and occlusion in input video.[5].

Author Sedakul combined the work of background subtraction and blob analysis to identify the object. [6]Haar feature based cascaded Adaboost classifier has been proposed for face detection. YOLO (You only look once) treats detection process as a regression problem to map the image in to object bounding boxes. In this, the input images divides in to grids, and each grid on put the bounding boxes on image. [7]

Fig.

3. Proposed work

The system is intended to identify number of vehicle on each traffic node with computer vision. System considers roads leaving a traffic signal as outgoing edge and roads coming towards a traffic signal as incoming edge. By considering number of waiting cars on road is may counting based on the segmentation process. Then mark the cars using Bounding box for count to open the signal based on cars count.In our proposed system, use HSV plane separation to get feature of each vehicle to create the dataset. Then, we have to use KNN classifier algorithm for classification. It will give the exact classification of each vehicle with exact result. The advantage is that we get exact result of seed and weed, HSV feature gives more accurate result. It is User friendly, simple process. Time delay will reduce.

Fig

4. Working

In this the traffic videos are given as inputs which is then converted into frames for further processing .The frames are then converted into HSV(Hue saturation value) images in which the planes are separated. We consider the saturation plane for binary conversion as it contains most of the information comparing the three planes. The binary conversion is done by fixing a threshold value, in which below threshold is taken as 0(Black) and above as 1 (White).

To get a proper image of a vehicle the holes are filled by KNN(K- Nearesr neighbor) classifier. Finally the bounding box is made and the vehicle count has been taken. This process is applied for all the four lanes and the density is measured. According to the priority the signals are changed in decreasing order respectively. Then the mat lab output is given to the hardware and verified.

Fig

5. Hardware implementation

Peripheral Interface Controller (PIC) is microcontroller developed by a Microchip, PIC microcontroller is fast and simple to implement the program when we contrast other microcontrollers like 8051. The ease of programming and simple to interfacing with


Page 308 of 310

other peripherals PIC become successful microcontroller.PIC mostly made up of Harvard architecture and also supports RISC (Reduced Instruction Set Computer), so it is faster than other microcontrollers. It works at 5V DC. Here we use PIC16f883 for implementation of traffic signals.

Fig

The SMPS (switched mode power supply) is used to convert the High voltage AC into low DC voltage. The transformer in SMPS step downs the AC voltage. TTL (Transistor-transistor logic) communicates between different voltages by balancing them. UART named CH34 is used for serial communication between the software and the hardware. The final results are further verified through LED’s.

6. Results

Case 1:

In case 1 we consider four lanes of traffic with various vehicles. According to the density of vehicles the traffic signals are varied with specific time delay. Comparing the four lanes B lane has more number of vehicles. Hence it has been considered as first priority. Later the other lane traffics are released according to their density.

Fig

Fig

Case 2:

In case 2 we consider four lanes of traffic with various vehicles including ambulance. The lane having ambulance is given as first priority. Later according to the density of vehicles the traffic signals are varied with specific time delay. Comparing the four lanes, ambulance is present in lane A. Hence it has been considered as first priority.Later the other lane traffics are released according to their density.

Fig

Fig


09 of 310

7. Conclusion

The main advantage of this project is four lanes are taken into consideration. We will get exact result of seed and weed, HSV feature gives more accurate result. It is User friendly, simple process and time delay will be reduced. In this a method for estimating the traffic and to prioritize the lanes in traffic signal using Image Processing is presented. It also detects the presence of ambulance and gives first priority to allow it. This is done by using the camera images captured in each lane. Each image is processed separately and the number of vehicles has been counted for each lane. Based on the number of vehicles, automatically the lane with high congestion will be allowed first to move. PIC microcontroller is also interfaced to demonstrate this process. The advantages of this new method include such benefits as use of image processing over sensors, low cost, easy setup and relatively good accuracy and speed. Because this method has been implemented using Image Processing and Mat lab software, production costs are low while achieving high speed and accuracy.Thus, it can be implemented in metropolitan city where there is heavy traffic all through the day.

Reference

[1] Bin Tian, Brendan Tran Morris, Ming Tang, Member, IEEE, Yuqiang Liu, Yanjie Yao, Chao Gou, DayongShen, and Shaohu Tang, Hierarchical and Networked Vehicle Surveillance in ITS: A Survey in IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS [2] NilakornSeenouvong ,UkritWatchareeruetai, ChaiwatNuthong, A Computer Vision Based Vehicle Detection and Counting System in 978-1-46783-8139-0/16/$31.00©2016IEEE [3] Ashwini B., Deepashree B., Dr.Yuvaraju B. N., Venugopala P. S., Identification of Vehicles in Traffic Video in International conference on Signal Processing, Communication, Power and Embedded System (SCOPES)-2016978-1- [4] TaqiTahmid , EklasHossain, Density Based Smart Traffic Control System Using Canny Edge Detection Algorithm for Congregating Traffic Information in 2017 3rd International Conference on Electrical Information and Communication Technology (EICT), 7-9 December 2017, Khulna, Bangladesh [5] T.SrideviKuruvaHarinathP.Swapna, automatic generation of traffic signal based on traffic volume in 2017 IEEE 7th International Advance Computing Conference [6] SedaKul, SüleymanEken, AhmetSayar, A Concise Review on Vehicle Detection and Classification in ICET2017, Antalya, Turkey.

[7] Asha C S, A V Narasimhadhan, Vehicle counting for traffic management system using YOLO and correlation filter in 978-1-5386-1112-8/18/$31.0©2018 IEEE0. [8] Ying Li1, Lingfei Ma1, Yuchun Huang, Jonathon Li Segment based traffic signal detection from mobile laser scanning data 978-1-5386-7150-4/18/$31.00 ©2018 IEEE [9] Tousif Osman, ShahreenShahjahan Psyche, J. M. ShafiFerdous, Hasan U. ZamanIntelligent Traffic Management System for Cross Section of Roads Using Computer Vision 978-1-5090-4228-9/17/$31.00 ©2017 IEEE


Page 310 of 310

intelligent transportation system by controlling traffic...

Documents