Vamsi Thulluri Krishna et al.

; International Journal of Advance Research, Ideas and Innovations in Technology

ISSN: 2454-132X
Impact factor: 4.295
(Volume 5, Issue 2)
Available online at: www.ijariit.com

Face recognition based door unlocking system using Raspberry Pi

Thulluri Krishna Vamsi Kanchana Charan Sai Vijayalakshmi M.
[email protected] [email protected] [email protected]
SRM Institute of Science and SRM Institute of Science and SRM Institute of Science and
Technology, Chennai, Tamil Nadu Technology, Chennai, Tamil Nadu Technology, Chennai, Tamil Nadu

ABSTRACT pass code to unlock the door. Although it provides security to

the doors, it also has its own drawbacks:
Today we are facing security issues in every aspect. So we
have to resolve these issues by using updated technology. In Firstly, if the system mainly uses a facial recognition module,
this project, we are using the Face recognition module to there might be a slight chance that sometimes the face may not
capture human images and to compare with stored database be detected and hence the door cannot be unlocked.
images. If it matches with the authorized user then the system Secondly, if the system uses a keypad to enter the pass code to
will unlock the door by an electromagnetic lock. The need for unlock the door, there might be a chance that the key maybe is
facial recognition system that is fast and accurate that recorded or can be observed by others without users consent.
continuously increasing which can detect intruders and Hence, two-step verification is developed which makes use of
restricts all unauthorized users from highly secured areas and facial recognition as the first step and pass code as its following
aids in minimizing human error. Face recognition is one of step. But the same issues pertain in the newly developed
the most Secured System than biometric pattern recognition system.
technique which is used in a large spectrum of applications.
The time and accuracy factor is considered about the major Thus, a new model which rectifies all the above issues is
problem which specifies the performance of automatic face developed.
recognition in real-time environments. Various solutions have
been proposed using multicore systems. By considering the 2. INTRODUCTION
present challenge, this provides the complete architectural Security deft has suggested various preferred approaches like
design and proposes an analysis for a real-time face biometric and password to enhance security. But the technology
recognition system with LBPH Algorithm. In this algorithm, is developed and growing with the usage of different
it converts the image from color to greyscale image and equipment's. The trend’s moved from fingerprint to face
divides into pixels and it will be allocated in a matrix form recognition. So we prefer a face recognition system for
and those images will be stored in the database. If an image is unlocking the door. Facial recognition is widely used in various
detected then microcontroller will send power to the motor industries and corporate sectors.
driver unit then the electromagnetic lock will unlock the door
and it will lock again when there is no power supply to that This door unlocking system mainly uses facial recognition. The
unit. Finally, this paper concludes for the advanced latest camera is used to detect the images and the images are
implementations achieved by integrating embedded system sending to the database. If the image matches with the admin's
models against the convention. image then the door is unlocked and an acknowledgement is
sent via Zigbee as “y” if it does not match the image an
acknowledgement is sent as “n”. A pass code column is shown
Keywords— Face recognition, Local binary histograms, which takes values from the keypad to unlock the door.
Keypad password, Electromagnetic lock
1. PROBLEM STATEMENT In today’s world by using smart devices we are make our needs
In the world of emerging technology, security became an smart. By following trends and updates we have to consider and
essential component in day to day life. Information theft, lack remove drawbacks in existing system and add more features
of security and violation of privacy etc. are the essential and updates. Face detection system is more complex because of
components which are needed to be protected. Using smart unstable characteristics. Example: let us consider glasses and
secure systems for door lock and unlocking became popular beard will show some impact to detect the faces. So by
nowadays. This is system is being adapted by many countries considering the different angles and multiple images of faces
and first grade countries such as USA, Japan etc. already makes and it will influence on detection process. The study of Open-
use of this system. This system provides either a facial CV and its inbuilt library functions helps to generate a code
recognition security feature or a keypad is provided to enter the will do correct and authentic facial recognition system with

© 2019, www.IJARIIT.com All Rights Reserved Page |1320

 Vamsi Thulluri Krishna et al.; International Journal of Advance Research, Ideas and Innovations in Technology
new and more efficient use of hardware. Human body will this approach we designed a new type of door locking and
identified as an input within environment by capturing live unlocking system.
video from a web camera and the process will be done on
captured video frames. The images will run through raspberry 4. SYSTEM ANALYSIS
pi3 and check with the stored data base, in this case, used an 8 4.1 System architecture
GB memory card. The compilation process will be performed In this, we are using Raspberry pi for connecting with camera
in VNC Viewer which helps to run Raspbian OS and the module to capture images continuously through Open-CV
response will send to the micro controller which is connected to platform and compared with stored data and send data to
Zigbee receiver and power is supplied to this micro controller another controller unit through Zigbee. Another Controller unit
by the transformer and a keyboard is connected to this micro receives through Zigbee and unlocks the door by using an
controller and display board is also connected to it. This will electromagnetic lock.
control the motor driver to lock and unlock the door. To run
this model there are different algorithms in that we took LBPH
because it will provide more accuracy results when compared Controlling Door lock
Web camera
to other algorithms. unit
Fig. 1: System architecture for proposed system
By this, we can say that door locking and unlocking by
detecting faces is a new model which includes an alternate 4.2 Transmitter
manual pass code unlocking system by using keypad which
helps to gain access to that door in necessary situations.

3. RELATED WORK
This section provides various approaches towards door
unlocking system. In previous works, they deal with different
algorithms technologies and equipment for unlocking the door.
In reference article 8 studies by Somjit Nath, Paramita Banerjee
proposed “Arduino Based Door Unlocking System with Real Fig. 2: Block diagram of a transmitter
Time Control”.This approach implemented with RFID codes to
scan for unlocking the door. So when a person wants to enter 4.3 Receiver
the door he needs to scan the card then he gains access to that
door. If he misplaces that card he cannot access through that
door and there is a chance of insecurity that anyone can access
to the door with that misplaced card. This stands as a drawback
to this system.

In reference article 7 studies by Charoen Vongchumyen,

Pakorn Watanachaturaporn, pattaya proposed “Door locking
system via web application”. In this approach, a web
application is designed to monitor the door so that user can
easily access door by his mobile and he can also check the
status whether it is locked or not. There is a drawback in this
system: when someone hacked and got security code then
hackers can easily access to that room.
Fig. 3: Block diagram of a receiver
In reference article 2 study by Suchit Shavi proposed ”Secured
Room Access Module”.In this approach a keyboard based door 5. EXISTING SYSTEM
unlocking system is implemented with a micro controller. So In the existing system, we are using password based door lock
the user needs to enter his password to unlock the door. In this system which makes the user memorize password all the time.
approach is secured when compared to the previous models and So to overcome this drawback we designing a new system with
well used in today’s world. Even though this system is secured face recognition.
there might be a drawback: If someone observes your
password, they can gain access to that door by using the same
password you have entered.

In reference article 1 study by Muhammad Kashif Shaikh,

Syedannas Bin Mazhar proposed “Comparative Analysis for a
Real Time Face Recognition System Using Raspberry Pi”. This
approach did an analysis of various algorithms on the face
recognition system. This analysis took LBPH (Local Binary
Pattern Histogram), Fisher Faces, Eigen Faces Algorithms for
comparison and checked with different processors to know time
complexity and accuracy of various algorithms. This approach
is most secured and without matching face no one can access
that door.

Hence this survey work proves that face recognition system is

the best to approach for using in present days by referencing Fig. 4: Block diagram of Secured room access system
© 2019, www.IJARIIT.com All Rights Reserved Page |1321
 Vamsi Thulluri Krishna et al.; International Journal of Advance Research, Ideas and Innovations in Technology
5.1 Flowchart
The flowchart operation of the micro controller of 8051
which is used in this module is shown in Fig 5. Until they
press the reset trigger, the operation which is performed first
and it asks for a new password and that will be stored in the
memory of the previous password. The module is then never
turn off while performing the operation. Then the micro
controller asks for a password every time there will be
someone who is trying to access the room. Then it will
compare the entered password with the previous password
which is set as original password. If the two passwords do
not match then it asks that person to re-enter the password, Fig. 6: Proposed modules Raspberry Pi with Webcam
when two passwords match then the door lock is released
and the door will unlock and it holds for some time and then
it will come back to position.

Fig. 7: Monitor runs with Raspbian OS

Fig. 5: Flow diagram for an existing system

5.2 8051 Microcontroller

8051 is the IC which is from C51 micro controller family. The
AT89S52 is a high performance and consumption is very less
the CMOS and Flash memory which is programmable and
erasable read only memory (PEROM) having 8-bit
microcomputer with 8Kb. The device is manufactured using
Atmel’s high density nonvolatile memory and it is compatible
with the industry standard of 80S51 and 80S52 instruction set
and pin out. The on-chip program memory is allowed by Flash
to be reprogrammed by a system or by a nonvolatile memory
programmer. By combining a versatile 8bit CPU with Flash on
a monolithic chip. The 8051 is a powerful microcomputer
which provides cost efficiency and highly flexible solutions too
many micro control applications.
5.3 Features
● It is an 8 bit micro controller
● It has 3 timers, one Serial port and four parallel Ports
● It is a 40 pin Dip package IC
● It has 128 bytes of RAM and 4KB of ROM.
● It operates with a frequency of 11.0592 MHZ
5.4 Disadvantages
● It Provides less security
● Every time password entering required for unlocking
● Micro controller cannot interface high power devices
directly

6. PROPOSED SYSTEM
In the new system, we are designing a door lock system with
the help of the face recognition module. In this, we are using
Raspberry pi which has many features that make the user
modify use in different smart applications. Fig. 8: Flow chart for the proposed system
© 2019, www.IJARIIT.com All Rights Reserved Page |1322
 Vamsi Thulluri Krishna et al.; International Journal of Advance Research, Ideas and Innovations in Technology
7. METHODOLOGY 8.1 Advantages
In this, we are using the LBPH (Local Binary Pattern  High reliability.
Histogram) Algorithm. This algorithm will give us more  It provides enough flexibility to suit the requirements.
accurate results when we compare to other types of algorithms  More secure due to face detection.
such as Fisher Face, Eigen Faces Algorithms in base paper.
This LBPH Algorithm will take number of images as you wish 8.2 Modules
in different angles and check those all images at the time of  Transmitter section: This consists of Raspberry pi, camera
face no recognition. In our case, we are taking 20 images of a and Zigbee module. The camera captures images through
person with different angles and it will be stored in our date Open CV and compared with stored authorized images. If it
base. For this algorithm, we are using VNC viewer to run matches the control will send some unique character through
raspbian os for detecting images from the data base. At first, we Zigbee.
have to save images by using data sets and after that, we will  Receiver section: This system consists of 8051 micro
train that faces to algorithm then it stores into the data base. At controller, keypad, LCD, Motor Driver, Zigbee and DC
first, it converts color images to gray scale images and then it motor. The controller receives data of unique character and
converts into pixels for detecting this will divides the image control movement of the motor. Here we are using L293D IC
into various pieces then it stores the values of each pixel. If for controlling direction.
pixels are less then it will be represented as 0 and pixels which
are high will be 1 then it will be arranged in 3 x 3 matrix format 8.3 Results
for recognizing the new images on screen compared to data All the modules present in Fig.1 are individually tested and
base stored images. Here are some different variations of faces later they integrated into a single main module. Raspberry Pi
that is capture. was programmed successfully using python with open CV and
Micro controller was programmed successfully using
Embedded C in Keil vision software to achieve face recognition
door unlocking system. The individuals who were authorized to
access the door he stands in front of the camera and it will
capture the image of that person and compare with the data
Fig. 9: Data set training images base images. If the images are recognized then automatically
door will unlock. When a person image is not detected then it
will ask for a manual pin to unlock the door, if that person
8. RESULTS AND DISCUSSIONS entered a valid pin the door will unlock. When a person entered
In this result we tested in different types of systems to know the the wrong pin then the door will not be opened and denied
accuracy of this system according to the system processing access to the room. Whereas when recognition of images, the
speed performance is depended. This case we captured and correct pin was detected, relay released the door and access was
shown different machine which is used in testing are listed granted for that person. This was tested with different
below with their configurations. combinations of faces and pins.
Table 1: Machine for testing accuracy
Operating Main 9. CONCLUSION
Machine Processor
system memory In this proposed door access system by using face recognition
Dell Core i the images are stored in a data base. This system is used door
Window 7 4GB
Inspiration (2nd gen) lock access for Residential and Commercial Purposes. Here we
Acer Aspire Corei5 have designed a highly secured door locking system by using
Window 7 4GB
(2nd gen) Raspberry pi
HP sleek book Corei3
Window 10 4GB
15 (4th gen) 10. REFERENCES
Acer Aspire Corei5 [1] (2017) “Comparative Analysis for a Real Time Face
Window 10 8GB
V5WE2 (4th gen) Recognition System Using Raspberry Pi” Muhammad
Kashif Shaikh, Syed Annas Bin Mazhar.
This is the graphical representation of the performance and time [2] (2017)”Secured Room Access Module” Suchit Shavi.
takes to detect the human face by using the LBPH Algorithm in [3] (2017) “Automatic Semantic Face Recognition”: Mark S.
above listed processors and systems. The system with high Nixon University of Southampton Southampton, United
configuration takes less time to recognize the face when Kingdom
compared to other processors. [4] (2017) “Real-Time Implementation of face recognition
system” by Neel Ramakant Borkar and Sonia Kuwelkar,
India
[5] (2017) “IoT based Home security through Digital Image
Process Algorithms” by A. Beatrice, Dr S. Britto Ramesh
Kumar and J. Jerlin Sharmila, India
[6] (2017) “Secured Room Access Module” by Suchit and
Shanvi, India
[7] (2017) “Door locking system via web application”
Charoen Vongchumyen, Watjanapong Kasemsiri,
Kiatnarong Tongprasert, Aranya Walairacht, Pattaya.
[8] (2016) “Arduino Based Door Unlocking System with
RealTime Control” Somjit Nath, Paramita Banerjee,
Rathindra Nath, Biswas, Swarup Kumar, Mitra.
Fig. 11: Graphical representation with time intervals
© 2019, www.IJARIIT.com All Rights Reserved Page |1323
 Vamsi Thulluri Krishna et al.; International Journal of Advance Research, Ideas and Innovations in Technology
[9] (2014) K.Gopalakrishnan, V.Sathish Kumar “embedded [11] (2012) “Face Recognition Based on Magnetic Door Lock
image capturing system using the raspberry pi system” System Using Microcontroller” Harnani Hassan, Raudah,
international Journal. Abu Bakar Ahmad Faculty of Electrical Engineering.
[10] (2014) “Development of Intelligent Automatic Door [12] (2005) “Real-time Embedded Face Recognition for Smart
System” Daiki Nishida, Kumiko Tsuzura1, Shunsuke Home” by F. Zuo and P. H. N. de.
Kudoh1, Kazuo Takai, Tatsuhiro Momodori. [13] (2000) “Automatic Door Opener” Pik-Yiu Chan, John D.
Enderle.

© 2019, www.IJARIIT.com All Rights Reserved Page |1324