IRJET - Smart Parking Guidance System

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 04 | Apr 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 1938
SMART PARKING GUIDANCE SYSTEM
Sathya Priya M. S1, Muthu Kumar.S2
1M.E. Student, Department of Electronics and Communication Engineering, SVCE, Sriperambadur
2Professor, Department of Electronics and Communication Engineering, SVCE, Sriperambadur
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - Considering the increase of urban,sub-urban
population and traffic congestion, smart parking is an
efficient and reliable strategy to overcome all the
parking related issues. The existing technique involves
the disposition of physical sensors in and around the
parking lot. The proposed system involves the technique
of Image Processing to overcome the major issues such
as computationefficiency,costmanagementandenables
real time parking methodology. The proposed system
consists of a camera which could acquire images from
the top view of parking space and the reason for using
the camera is that it could sense the entire parkinglotat
once. The captured images are treated as frames and
processed to exactly know the vacant space detection.
Meanwhile the vacant spots are assigned a unique
number tag and the driver is provided with the
information of unoccupied slots. An automatic parking
system is a less complex method for parking vehicles for
both the drivers and administrators
Key Words: Digital Image Processing,Matlab,Video
Dataset, Computer Vision.
1. INTRODUCTION
Nowadays car has become an essential luxury goodfor
the people. Though it has made a significant impact in
people’s life, the problem of traffic jam is inevitable.
The misfortune contributed by the traffic jam is hardly
avoidable. The concept of urbanization has made a
negative impact on the quality of people’s life. The
wastage of fuel,roadaccidents andemissionofharmful
gases such as carbon-di-oxide has led tomiscellaneous
innovations in researcher’s perspective. Plenty of
research works are contributed solely in the field of
efficient ways of transportation and the methods to
overcome the issues faced in the particular sector.
Summarizing all the issues, we have designed a smart
parking guidancesystemwhich couldassistthedrivers
to park their respective vehicles in a particular lane
without interference to the adjacent cars. The driver
does not know the empty parking slot when he enters
the parking area. So, inorder to aid him throughout the
complete process of parking, few cameras are fixed at
the top angle to direct him towards the process of
parking the car. Here we have used a video dataset to
locate the availability of vacant slots. The concept of
Image Processing [1] has been utilized in the entire
research work and the Computer Vision algorithm has
added a major value to it.
The system provided not only gives the exact location
of unoccupied slot but also assigns a unique numberto
the particular lot which could provide more
information to the driver and the driver could park his
vehicle in the dedicated slot provided to him.
2. RELATED WORK
i. Sensor Based Methods
The sensor based methods [2], [3], [4] use the
deployment of physical sensors and it could cost more
money which is a very potential problem to be
addressed. The sensors could make the work very
easier but the sensors (forexample,ultrasonicsensors,
IR sensor) should be kept in andaroundtheparkinglot
which is a very tedious process andabitexpensive.For
example, the ultrasonic sensor could accurately
calculate the distance and tell us the vacant space
where the car needed to be parked. Many kinds of
sensor technologies have been discussed [5] which
could be installed on the grounds where the car is
about to be parked, in and around the parking lot.
ii. RFID tag based Intelligent Parking Assistant
The Intelligent Parking System (IPA) [6] aimed at
mitigating current public parking management
problems. The architecture deals with the on-street
parking availability and allows the driver to reserve a
convenient parking space. When a car enters or leaves
the particular parking area the RFID reader and
magnetic loop detect the action and it automatically
sends the information regardingthecarparkingstatus.
This system is applicable only to a minimized area and
it could not cover an entire parking area. This is the
major disadvantage of this RFID system which is been
incorporated in the process of Intelligent Parking
Assistant.

iii. VANET based parking reservation system
Vehicular Adhoc Network(VANET) [7] is a kind of
networkcreatedinAdhocmannerwheredifferentkind
of vehicles exchange the information with each other
vehicle over a wireless medium. It aims to mitigate the
cost of installing physical sensors but it requires a
special kind of equipment to be installed in cars and
also along the roadside. It requires more cost, labour
and time and this kind of system could not be
realistically implemented. Certain kind of intelligent
transportation system aims to reserve a parking spot
prior to their trip. Instead of introducing dynamic
message signs which could update the available
parking space, these schemes introduce many
optimization strategies. The proposedworkdealswith
the usage of camera alone for the entire parking
reservation system.
3. PROPOSED METHOD
In this system the video footage of the entire parking
area is captured via a camera. The video is acquired
from the top view of the building at a particular angle
so that the entire parking space could be covered at
once. The cameracouldalsobethesurveillancecamera
fixed for the security purpose thereby it is made sure
that the cost involved in this system is considerably
reduced without installing various other sensors. The
video is captured and the captured video is segmented
into many frames. The video dataset we have, consists
of 1400 frames and out of these 1400 frames, a key
frame is selected at an interval of every twenty frames
to reduce the computational complexity.
The pre-processing phase is performed by the
conversion of captured RGB image into gray scale
image and unwanted noise is eliminated using median
filters besides improving the algorithm and efficiency
rate thus enhancing the image frame. The features are
extracted using key pointsfromdifferentregionsofthe
image frame using SpeededUpRobustFeatures(SURF)
based feature extraction method and the features are
matched based on a thresholdvaluetonotifythedriver
whether the slot is vacant or full. Meanwhiletheempty
slot number is informed to the driver based on the
threshold value fixed with respect to each and every
slots. Fig -1 shows the basic flow of the system.
Fig -1: Basic flow of the system
4. SOFTWARE TOOL REQUIRED
We use MATLAB image processing components here.
MATLAB is a high level language and serves as an
interactive environment for various numerical
computation and programming analysis.Itcanbeused
to analyze miscellaneous data, develop powerful
algorithms and create models. Besides Image
processing MATLAB is also used in different domains
like signal processing, video processing etc.
4.1 OPERATIONS IN MATLAB
The prediction of vacant slots and occupied slots could
be determined by the operations performed in the
MATLAB: -
 The RGB image is converted into grey scale
image for easier computation.
Parking slot video input
Sequence of image frames
Pre-processing
Enhancement of the frame
SURF based feature extraction
Feature Matching
Display of vacant slot number

 Extractionofkeypointsanddescriptorsare
taken into account for feature extraction.
5. SYSTEM IMPLEMENTATION
A. Input Video Frame Selection
The video dataset is segmented into 1400 images or
frames and a key frame is selected at an interval of
nearly 20 frames for easier computation. These frames
are given as an input for processing. These images are
captured bycamera which senses the real time parking
space. Fig -2, Fig-3 andFig- 4 are the examples ofsome
key input frames that we have used in our project
which are selected for processing and computation.
After selectingthe keyframes,theframesaresubjected
tovariousprocessingwhicharediscussedinthefurther
subdivisions.
Fig -2: Sample input frame 1
B. Pre-Processing and Enhancement
The input frame selected is read and the RGB image
frame is converted into gray-scale image. The
rgb2gray(RGB)convertstheoriginalthreedimensional
image to gray image. This is done by eliminating both
the hue and saturation while retaining only the
luminance. The RGB image has to be first converted to
gray-scale image in order to avoid coding complexities
andtofacilitateeasiercomputations.Thepre-processed
image is enhanced using median filters to remove the
unwanted noise.
C. SURF Based Feature Extraction
The detectionof a particularparkingspaceneedsmuch
effort. But with the good parking cameras and with
wide angle of view, the parking space detection could
be approached with ease. The edges of each and every
boundaryofaparkinglotisclassifiedviaedgedetection
technique using blob detection method. SURF is a
detector-descriptor scheme. Many feature extraction
techniques are known such as Scale-Invariant Feature
Transform(SIFT), Speeded-UpRobustFeatures(SURF)
etc. It is proved that SURF has outperformed SIFT in
feature extraction and matching. Its fast performance
originates from a detection stage of interest points,
where the detector uses a scale invariant blob detector
based on a mathematical determinant. With the help of
Hessian matrix, the determinant is calculated. This
collection of data that relies on the Hessian matrix can
later be used for both scale selection and location
placement with the help of a set of box filters and the
usage of integral images, the detector can approximate
the second order Gaussian derivatives. This approach
can be mathematically described, where the input
image is (x, y) and S stores the sum of all pixel within a
rectangular region and is explained by the below
equation: -

(1)
As described before, SURF utilizes a blob detector to
find so called keypoints in given image. The algorithm
is using the Hessian matrix, which is used to compute
the local maxima around the chosen points. The given
Equation (2) below represents the determinant
operation, where L is the local maxima, ρ illustrates
the coordinates (x, y) and σ are chosen as the scale in
given image and H is the hessian matrix.
(2)
D.Feature Matching
The most commonwaytofindasimilaritybetweentwo
descriptors say Aand Bis to use a Brute Force Matcher.
The reference frame and the current frame are
simultaneously compared to find the similarities
between two frames. The aim of the Brute Force
Matcher is to find the similarity of features and is
performed by calculating Euclidean distance which is
represented bytheequation(3)wherexrepresentsthe
128-dimension vector, n could be an integer and xn is
the distance between points A and B.
(3)
These all are achieved bycomputer visionalgorithm in
which a camera senses an image as the human could
perceive an image.
EMPTY SLOT IDENTIFICATION
The empty slot is identified when a car goes out of its
respective slot. The empty slots are found by fixing a
threshold value with the help of intensity values
between the previous and current frame. Based on the
threshold value between previous and current frames,
the empty slots are continuously updated to the user
identified with ease. The vacant slots are numbered
with a unique number which provides an easy
accessibility to the user for identification of parking
lots. The vacant spaces are displayed on the Matlab
console. Fig- 5 and Fig-6 depictsparkinglotnumbering
and vacant spaces display for a sample frame.
Fig -5: Parking Lot Identification
Fig -6: Vacant Space Display
6. CONCLUSION
Animagebasedmethodofdetectingtheavailabilityofa
car park was modelled and tested with different
occupancy scenarios of car parks using MATLAB. The
method of analyzing an aerial view of the car park has
been presented step by step. This method consists of
converting the RGB image to grey for simple analysis,
finding car park coordinates from a parking space
thereby removing noise and determining whether car
parks are vacant or filled. The concept behind the work
is to discover the parkingsystemsolelybasedonimage
processing rather than introducing costly physical
sensors. Intelligent parking system is developed using
the concept of image processing thereby reducing the
cost of miscellaneoussensorsandwiringhassle.Future
research will be focused on machine learning and
artificialintelligencealgorithmsasacomplementofthe
intelligent parking system.

FUTURE SCOPE
Computer Vision has achieved its heights of peak with
the advancement of deep learning algorithms
specifically convolutional neural networks (CNN).
CNNs have the ability to classify even the dense pixel
segmentation. From the perspective of automated car
parking, CNN could be considered as a glimpse into
next generation and cameras could play a majorrolein
future automated parking systems
REFERENCES
[1] Paula Tatulea, Florina Calin,Remus Brad,Lucian
Brancovean and Mircea Greavu, ”An Image
Feature-Based Method For Parking Lot
Occupancy”, Future internet ,2019.
[2] Fabio Duarte and Carlo Ratti, “The impact of
autonomous vehicles on cities”, Journal of Urban
Technology, pp. 1–16, 2018.
[3] Muhammad Alam,Davide Moroni, Gabriele
Pieri,Marco Tampucci,MiguelGoes,JoseFonseca,
Joaquim Ferreira and Giuseppe Riccardo Leone,
“Real-Time Smart Parking Systems Integration in
Distributed ITS for SmartCities,JournalofAdvance
Transportation, Wiley Publications,2018.
[4] Rachapol Lookmuang, Krit Nambut, Sasiporn
Usanavasin,” Smart Parking usingIOTtechnology”,
International Conference on Business and
Industrial Research(ICBIR),2018.
[5] Trista Lin, Herve Rivano and Frederic Le Mouel ,’A
Survey of Smart Parking Solutions”, IEEE
TransactionsonIntelligentTransportationSystems
Vol 18, NO 12,2017.
[6] LucaMainetti, Palano, L. Patrono, M.L.Stefanizzi,
R.Vergallo “Integration of RFID and WSN
technologies in a Smart Parking System”, IEEE
International Conference on Software,
Telecommunications and Computer Networks,
2014.
[7] Azizur Rahim, Feng Xia, Xiangjie Kong, Zhaolong
Ning, Noor Ullah, Jinzhong Wang, Sajal K.Das,
“Vehicular Social Networks:A survey”, Pervasive
and Mobile Computing ,Elsevier,2017.
[8] Bill Yang Cai,Ricardo Alvarez,Michelle Sit,Fabio
Duarte,Carlo Ratti ,”Deep Learning Based Video
System for Accurate and Real Time Parking
Measurement”,IEEE Internet of Things
Journal:special issue on enabling a smart
city:Internet of Things Meets AI ,2019.

IRJET - Smart Parking Guidance System

More Related Content

What's hot (20)

Similar to IRJET - Smart Parking Guidance System (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET - Smart Parking Guidance System