Special Issue - 2020 International Journal of Engineering Research & Technology (IJERT)

ISSN: 2278-0181
NCICCT - 2020 Conference Proceedings

Raspberry Pi based Face Recognition System

Dr. J. Preetha1, M. Manirathnam2, A. Chaitanya3, R. Prakash Raj4
1
Professor, Department of Computer Science and Engineering
2,3,4
Student of Computer Science Engineering,
Muthayammal Engineering College(Autonomous), Rasipuram, India.

Abstract-- Nowadays, computers and smart phones have to be increased . The face recognition system consists of
significantly contributed our daily life where numerous two important steps, the feature extraction and the
computations and adjustments are being accomplished by classification.
such technologies. Securing homes has become one of the Face recognition has a challenge to perform in real
concerning issues that facing many people. With the expanded
duration of leaving the home, homes are being more
time. Raw face image may consume a long time to
vulnerable for several threats especially being burgled. Apart recognize since it go through from a huge amount of pixels.
from the threats, there are different cases where securing or One needs to reduce the amounts of pixels. This is called
monitoring the house is very critical such as the existence of dimensionality reduction or feature extraction, to save time
elderly individuals or kids with baby sitter. For this manner, for the decision step. Feature extraction refers to convert
home security system or so-called Home OS has been face space into a feature space. In the feature space, the
proposed in order to provide more secure arrangements. Such face database is represented by a reduced number of
concept aims to turn the home into a smart in which different features that retain most of the important information of the
tasks especially monitoring can be performed remotely. original faces.
Monitoring and controlling some tasks inside the house would
have the ability to provide maximum safety.
The most popular method to achieve this target is
Recently, electronic door lock systems are one of the most through applying the Eigenfaces algorithm. The Eigenfaces
popular security systems that is being installed for many algorithm is a classical statistical method by using the
residents and business places. The key characteristic behind linear KarhumenLoeve transformation (KLT) also known
such systems lies on the reliability in which the authorized as Principal component analysis. The PCA calculates the
individuals can gain the permission to access the doors eigenvectors of the covariance matrix of the input face
throughout a secure system that has an interactive interface. space. These eigenvectors define a new face space where
Such analysis takes different forms such as analyzing facial the images are represented. In contrast to linear PCA, N-
image or video stream. The main goal is to develop a PCA has been developed.
prototype thathas the ability to simulate the wireless tasks
including monitoring and controlling digital door lock. Such
system would have the capability to provide secure II. LITERATURE SURVEY
monitoring of houses. Current scenario is that we manually go and open
the door for any visitor and if we are not present when
I. INTRODUCTION some our family member comes to our home this may lead
Machine learning (ML) is the scientific study of them wait until we reach. To provide the good facility for
algorithms and statistical models that computer systems use the relative we provide them access to the doors even when
to perform a specific task without using explicit we are absent, this may help to better experience for the
instructions, relying on patterns and inference instead. It is visitors.
seen as a subset of artificial intelligence. Machine learning The major disadvantage is that appearance of n-
algorithms build a mathematical model based on sample number of people at same time hard to detect the existed
data, known as "training data", in order to make predictions faces, the detection of face need some time to provide the
or decisions without being explicitly programmed to access to the trained persons, the movement of person may
perform the task. Machine learning algorithms are used in a result into wrong detection.
wide variety of applications, such as email filtering and So, we provide a machine learning approach which
computer vision, where it is difficult or infeasible to will be used to scan those who all are relatives or family.
develop a conventional algorithm for effectively Then we provide the access to the detected faces by
performing the task. comparing with databases and we also provide the training
Face recognition systems are now providing the need process for new friends and salesmen.
for security to cope up with the current misdeeds. It is The trained model provides faster results when
really influential with the market information that compared to the manual methods.The results are more
undoubtedly depicts the rising fascination of the face accurate due to the supervised training of the detection
recognition system. In the present era, the threat of model.The proposed system will reduce the human efforts
protecting the information or physical property is becoming and make sure no one enter the house without access and
more and more difficult and important. Now a day the minimal possibilities of error.
crimes of computer hackings, credit card fraud or security
violation in a company or government building has noticed

Volume 8, Issue 08 Published by, www.ijert.org 1

Special Issue - 2020 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCICCT - 2020 Conference Proceedings

III. IMPLEMENTATION DETAILS

A pattern recognition task performed exclusively on
faces is termed as face recognition. It can be described as
classifying a face either known or unknown, after matching
it with stored known individuals in a database. It is also
advantageous to have a system that has the capability of
learning to identify unknown faces. The outline of typical
face recognition system implemented specifically for N-
PCA.
The working methodology is divided into 3 parts,
namely Camera module, Raspberry Pi module and
Electronic bolt lock as shown in figure 1.

Figure 2 Flow Chart Model

The phase 2 is the implementation of face

recognition algorithm. The internal design of the proposed
system is shown in figure 3. The Face Recognition system
includes: An automated mechanism that scans and
captures adigital or an analog image /of living personal
characteristics (Enrollment module). Another entity which
handles compression, processing,storage and compression
Figure 1 Hardware Architecture of the captured data with stored data (Database). The third
interfaces with the application system(Identification
A. Face recognition: module).
The implementation is executed in two phases.
The phase 1 is related to saving the face portion. The flow
chart shown in figure 2 explains the working of the module
which uses the Raspberry Pi. Initially, the frame is taken
and the face is detected. To capture the images, we are
using open cv with raspberry Pi. After detecting the face,
the process of extraction is done along with resizing the
image. Then, the process of saving the extracted image is
carried out. After saving the image the real time training is
done, in which the respective names of the images are
trained for correct analysis of the person. Finally, the faces
of person will be saved in folder with the names of the
persons. Those faces will then be used to recognize face.

Figure 3 System Architecture

The different Modules of Face Recognition

System are:
• User Interface: User interface captures the analog or
digital image of the person's face. In this we take number
of photos of a single person either in analog or in a digital
form.

Volume 8, Issue 08 Published by, www.ijert.org 2

Special Issue - 2020 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCICCT - 2020 Conference Proceedings

• Enrollment Module: In the enrollment module the eachpixelwithitsneighbourhood.Takeapixelascentreand

obtained sample is preprocessed and analyzed. This thresholditsneighboursagainst.Iftheintensityofthecentre
analyzed data is stored in the database for the purpose of pixelisgreater-equalitsneighbour,thendenoteitwith1and 0
future comparison. if not. You’ll end up with a binary number for each pixel,
• Database: The database compresses the obtained sample justlike11001111.Sowith8surroundingpixels,you’llend up
and stores it properly. It should have retrieval property with 2^8 possible combinations, called Local Binary
also so that it compares all the stored sample with the PatternsorsometimesreferredtoasLBPcodes.ThefirstLBP
newly obtained sample and retrieves the matched sample operator described in literature actually used a fixed 3 x 3
for the purpose of verification of the user and determine neighbourhood just likethis.
whether the match declared is right or wrong.
• Verification: The verification module also consists of a IV. TESTING PROCEDURE AND RESULT
pre-processing system. Verification means the system The proposed system works are as follows:
checks as to who the person says he or she is and gives a • Interface the camera to capture live faceimages.
yes or no decision. In this module the newly obtained • Create a database of an authorized person by using
sample is compared and pre-processed with the sample Graphical user interface shown in belowimage.
stored in the database.
B. Text recognition
Text recognition in images is a research area which
attempts to develop a computer system with the ability to
automatically read the text from images. These days there
is a huge demand in storing the information available in
paper documents format in to a computer storage disk and
then later reusing this information by searching process.
One simple way to store information from these paper
documents in to computer system is to first scan the
documents andthen store them as images. But to reuse this
information it is very difficult to read the individual
contents and searching the contents form these documents • Write name in the blank block in GUI model to store
line-by-line and word-by-word. the personimageontheirnameinthedatabase.Andthenclick
The challenges involved in this the font characteristics onstart.
of the characters in paper documents and quality of images.
Due to these challenges,computer is unable to recognize the • After click on start one video frame window will
characters while reading them. Thus there is a need of open. It captures the images and creates thedatabase.
character recognition mechanisms to perform Document • After competing capturing the images click onTrain.
Image Analysis (DIA) which transforms documents in (Above two process images are following shown
paper format to electronic format. In this paper we have respectively.)
discuss method for text recognition from images. The
objective of this paper is to recognition of text from image
for better understanding of the reader by using particular
sequence of different processing module.
C. Software and Techniques for Face Recognition
This System is used Python 3.5+ software and python
programming.WithOpenCV(OpenSourceComputerVision)
isalibraryofprogrammingfunctionsmainlyaimedatreal- time
computervision.
• Haar CascadeClassifier
ObjectDetectionusingHaarfeature-
basedcascadeclassifiers is an effective object detection
method proposed by Paul Viola and Michael Jones in
their paper[5], "Rapid Object Detection using a Boosted
Cascade of Simple Features" in 2001. It is a machine • After click on start one video frame window will
learning based approach where a cascade function is open. It captures the images and creates thedatabase.
trained from a lot of positive and negative images. It is • Finally, it checks with the database to provide the
then used to detect objects in other images. access for the peoples.
• Local Binary Patterns HistogramsTechnique
The Local Binary Patterns methodology has its
roots in 2D
textureanalysis.ThebasicideaofLocalBinaryPatternsisto
summarize the local structure in an image by comparing

Volume 8, Issue 08 Published by, www.ijert.org 3

Special Issue - 2020 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
NCICCT - 2020 Conference Proceedings

[5] LeratoMasupha,TranosZuva,SelemanNgwira,OmobayoEsan,
"Face Recognition Techniques, their Advantages,
Disadvantages and Performance Evaluation", IEEE2015.
[6] Paul Viola, Michael J. Jones, "Robust Real- Time
Face Detection",InternationalJournalofComputerVision57(2),
2004.
[7] Ayushi Gupta, Ekta Sharma, NehaSachan, and Neha Tiwari,
"Door Lock System through Face RecognitionUsing
MATLAB", International Journal of Scientific Research in
Computer Science and Engineering, Vol-1, Issue-3, 30 June
2013.

V. CONCLUSION
It can be concluded that automated door lock system in
home using human face recognition technique works quite
well. Certainly, it can be improved for yielding a better
result particularly by paying attention in feature extraction
or recognition process. This improvement may help the
recognition process become more robust. The success rate
of the proposed system in recognizing facial images of the
person who are standing in front of the door is about 82%.
Also, face recognition systems have addedvalue to security
and surveillance areas such as airports, schools, company,
corporation etc.

VI. REFERENCE
[1] HteikHtarLwin, Aung SoeKhaing, HlaMyoTun, “Automatic
Door Access System Using Face Recognition”, International
Journal of Scientific & Technology Research Volume 4, Issue
06, June2015.
[2] Samuel Lukas, Aditya Rama Mitra, RirinIkanaDesanti, Dion
Krisnadi "Student Attendance System in Classroom Using
Face Recognition Technique" IEEE2016.
[3] Ratnawati Ibrahim, "Study of Automated Face Recognition
System for Office Door Access Control Application",
IEEE2011.
[4] Mrs. J. Preetha and Dr S.Selvarajan” Computer aided
diagnostic system for automatic cervical herniation
classification” Journal of medical imaging and health
informatics , volume 6, ISSN: 2156-7018, 2156-7026 2016

Volume 8, Issue 08 Published by, www.ijert.org 4