International Journal of Research in Engineering, Science and Management 432

Volume-3, Issue-8, August-2020

journals.resaim.com/ijresm | ISSN (Online): 2581-5792 | RESAIM Publishing

Night Vision Security Patrolling Robot Using

Raspberry Pi
J. N. Amrutha1*, K. R. Rekha2
1
M.Tech. Student, Department of Electronics and Communication Engineering, SJB Institute of Technology,
Bangalore, India
2
Professor, Department of Electronics and Communication Engineering, SJB Institute of Technology,
Bangalore, India
*Corresponding author: [email protected]

Abstract: The project deals with the use of Raspberry pi to also available. It is not remote controlled and just serves as
survey Smart monitoring system. These days’ video surveillance is another collection of ground-level cameras and an alarm. The
important in terms of security. High end cameras are needed in Security patrolling autonomous robot presented in this report
commercial spaces, schools and hospitals, warehouses and other
demanding indoor and outdoor environments. Current will combine several technologies to feature a robot with
technologies require expensive RFIDs and hence the security capabilities for detection and authentication. It will also act as
domain becomes expensive in all, so work on this was needed. The a preliminary alarm device, a surveillance camera, and function
project explains the use of Raspberry Pi, a low cost single onboard in an indoor / outdoor environment. It can be found in office
computer. This new technology is less expensive, and is used as a buildings, shopping centers or even larger homes. Raspberry Pi
standalone image processing platform in this project. It makes is a computer sized to a card.
more use of mobile technology. Security systems currently need
tons of costly components and a complicated installation process. It works almost identical to a computer. Various types of
There are currently two main types of systems available. The first surveillance systems are available, such as monitor, CCTV, etc.
is that of a wired machine. One drawback is that installing a wired In these types of surveillance systems, the person who is
system can take a considerable amount of time and money. stationary and situated in that area can only see what is
Another drawback is that the establishments a permanent part of happening in that location. And here, even though the user
that. When the owner moves the machine must stay in place. travels from one location to another, he / she can keep track at
Keywords: Raspberry Pi. the exact moment of what is going on in that particular place.
Also, it provides anonymity on both sides, is another plus.
1. Introduction Raspbian OS is the operating system which is used here.
There are a range of techniques in a wide variety of images
Security is a vital asset for lot of companies and homeowners.
that can effectively detect upright frontal faces. Such systems
It also avoids damage done to individuals and avoids stolen or
can tackle non-upright face detection directly. This program
destroyed personal belongings. Each year, companies and
explains progress towards a device that can detect and
homeowners set aside a large budget only for security
recognize faces accurately and in real time, regardless of
measures. On average a security guard's annual salary is around
position, based on hair-like characteristics. Viola et aland
$25,000. It expense increases with the number of alarm
introduces hair Raspbian OS must be mounted in order to send
systems, surveillance cameras and hired security guards. In
the picture to the mobile Closed circuit television monitoring
2011 homeowners spent about $20.64 billion on united state
system has now become an essential technology in today's
home security systems. These statistics show how imperative
society. Robots in our lives have found a significantly growing
safety is for companies and homeowners alike.
market for a variety of jobs. Their use in the military and other
Alongside security guards, there are a variety of
security sectors is increasing day after day, like features,
technologies that help to ensure security. Security Camera is
improved by Lienhart et al. The detection technique is based on
one of the most common types of surveillance. This security
the wavelet template idea which defines an object's shape in
method does not prevent the crime, but helps to identify
terms of a subset of image wavelet coefficients. Given its
criminals to law enforcement when a crime occurs. There are
computational efficiency and simplicity, this approach is
also security systems, where a security may sound once
chosen.
triggered to prevent offenders from further breaking in. These
systems typically send a warning to law enforcement or a
2. Fundamentals of the Project
security station, and can be triggered within a certain range by
sound, movement. Remote controlled robot of rover form is The proposed system, Raspberry pi is installed with the night
vision camera that helps the system find the intruders and go for
 International Journal of Research in Engineering, Science and Management 433
Volume-3, Issue-8, August-2020
journals.resaim.com/ijresm | ISSN (Online): 2581-5792 | RESAIM Publishing

the automation. The robot covers a specific area and checks for particles that carry electricity around the circuit).
any intruders when an intruder is detected, the owner is alerted
C. DC Motor
by the buzzer sound.
A DC motor is one of a class of rotary electrical machines
which converts electrical direct current into mechanical energy.
The most that types of magnetic fields depend on the forces
produced. Nearly all types of DC motors have some internal
function, electro mechanical or electronic, to adjust the
direction of current flow within part of the motor periodically.
DC motors were the first type widely used, as they could be
powered from existing distribution systems for direct current
lighting. The speed of a DC motor can be regulated over a wide
range, either by using a variable supply voltage or by adjusting
the current force in its field windings. Small DC motors are used
in tools, appliances and toys.
D. PIR Sensor
PIR infrared sensor module operates on operational infrared
sight, programmed control options. It has high dependability,
low power consumption, sensitivity variable, task mode low
Fig. 1. Functional block diagram voltage. Generally used as part of various electrical gears for
automated detection, especially for battery-powered and
We propose a robot patrolling security which uses night programmed control systems.
vision camera to secure any premises. The robotic vehicle E. IR LED
moves at specific intervals, and is equipped with camera and
An IR LED (infrared light emitting diode) is a solid state
sound sensors for night vision. In patrolling it uses a predefined
lighting device (SSL) that emits light within the
line to follow its path. It stops at specific points and if sound is
electromagnetic radiation spectrum infra-red range. Infrared
detected moves to next points. To patrol assigned area, the
light, which is electromagnetic radiation in the range of 700 nm
system uses IR-based path following system. It monitors each
to 1 mm, is produced cheaply and efficiently with IR LEDs. IR
area using 360degree rotating HD camera to detect any
LEDs are useful in different types of electronics, including
intrusion. It is capable of tracking sound at the premises. Any
many types of television and other electronic remote controls.
sound after company is closed and its predefined path begins to
Used with infrared cameras, IR LEDs will serve as a spot light
move towards the sound. It then scans the area using its camera
while remaining naked-eye invisible. Infrared light, which is
to detect any identified human faces It captures and begins to
electromagnetic radiation in the range of 700 nm to 1 mm, is
transmit images of the situation immediately upon detection of
produced cheaply and efficiently with IR LEDs. IR LEDs are
sound or human face. Here we use IOT Local Area Network
useful in a number of electronic types including many types of
(LAN) to get transmitted images and display them with warning
television and other electronic remote controls. IR LEDs used
sounds to the user. So we're putting forward a fully autonomous
with infrared cameras can act as a spot light while remaining
security robot that runs tirelessly and patrols large areas alone
invisible to the naked eye. Since IR LEDs can be used in
to secure the facility.
conjunction with a variety of different sensor types, they are
A. Raspberry Pi 3 becoming popular in Machine-to-Machine (M2M) and Internet
The Raspberry Pi is a series of small single-board computers of Things (IoT) environments.
produced in the UK. The original model became much more F. TIP 120 Transistor
popular than expected, selling for uses such as robotics outside
The TIP 120 is an NPN darlington transistor with a current
its target market. It does not include peripherals (e.g. keyboards,
gain of 1000 which is a good choice for interfacing to an
mice and cases). However, some of the accessories have been
Arduino or other microcontroller with many higher current or
included in a number of official and non-official Packages.
higher voltage loads. It can be used for controlling
B. Night Vision Camera microcontroller DC motors, solenoids, or LED strings with
Dim light from the night scene enters the front of the lens. minimum current draw
The light is made of photons (light particles) of all colors. As G. LCD Display
the photons enter the camera, they hit a light-sensitive surface
LCD (liquid crystal display) is the technology used in
called a photocathode. It's a bit like a very precise solar panel
notebook displays and in other smaller computers. Like light
whose job is to convert photons into electrons (small, subatomic
emitting diode (LED) and gas plasma technologies, LCDs allow
 International Journal of Research in Engineering, Science and Management 434
Volume-3, Issue-8, August-2020
journals.resaim.com/ijresm | ISSN (Online): 2581-5792 | RESAIM Publishing

displays to be much thinner than cathode ray tube (CRT) paper using the nice ARM9 board support package (BSP)
technology. LCDs consume much less power than LEDs and S3C2440. This application device captures video, shares
gas displays because they operate on the principle of blocking between networked devices, and also alerts the controller with
light rather than emitting it. An LCD is either made with a a short message service warning, as the client needs. This
passive matrix or with an active display grid for the matrix. A system operates in an environment in real time, and is supported
thin film transistor (TFT) display is also known as the active by RT Linux embedded. This system provides low cost and
matrix LCD. The passive LCD matrix has a grid of conducers high efficiency intelligent monitoring systems with low power
with pixels located in the grid at each intersection. To monitor consumption, such as elevators, home security systems, etc.
the light for any pixel, a current is sent on the grid between two Power consumption small. This real-time system provides a
conducers. An active matrix has a transistor located at each client video monitor with the help of friendly ARM9 BSP,
intersection of pixels which requires less current to control a unlike other embedded systems.
pixel's luminance. For this reason, the current in an active [2] In a wide variety of images, there are a number of
matrix display can be turned on and off more often, increasing techniques that can successfully detect upright frontal faces [1].
the refresh time on the screen (for example, the mouse may Some systems can explicitly address non-upright face detection
appear to be traveling faster around the screen). [3]. This paper describes progress towards a system that can
detect and recognize faces on the basis of haar-like features,
H. Sound Sensor
regardless of pose reliably and in real-time. In a wide variety of
A sound sensor is defined as a module that detects and images, there are a number of techniques that can successfully
converts sound waves into electrical signals via its intensity. detect frontal upright faces [1]. Some systems can explicitly
Before we move on to how to integrate a sound sensor with address non-upright face detection [3]. This paper describes
Arduino, let's first take a look at the tutorial 's sound sensor progress towards a system based on hair-like features that can
module that we will be using. Based on the LM386 power detect and recognize faces irrespective of pose reliably and in
amplifier, the Grove – sound sensor module is a simple, low real-time. We have found that the simple attempt-all-poses
power and highly compatible option for easy kick-starting of system actually yields higher operating characteristics of the
your next sound sensing project. With a wide voltage range and receiver (ROC) curve, although it is slower. Due to its
the potentiometer adjustable output, it is readily capable of computational efficiency and simplicity this approach is
detecting the ambient sound strength. chosen.
I. Wi-Fi Module [3] Multi-environment robot for surveillance and live
It uses a 32-bit RISC CPU based on the 80 MHz (or streaming is designed to install real-time surveillance system
overclocked to 160 MHz) Tensilica Xtensa L106 running at. It possible within a local network. The live streaming is
has a boot ROM of 64 KB, RAM instruction of 64 KB, and accomplished using mpeg streamer and the server-client model
RAM data of 96 KB. External flash memory is accessible is built using java as IP-based deployment offers access from
through SPI. The ESP8266 module is a low cost standalone image quality and is also advantageous in terms of scalability
wireless transceiver which can be used to develop IoT and versatility. But IP-based systems require some networking
endpoints. Microcontroller will need to use set of AT knowledge and these systems are too expensive than the analog
commands to communicate with the ESP8266 module. Using ones A server-client model incorporates this robot controlled
UART, the microcontroller communicates with the module with raspberry pi. This client-server model is built on java, and
ESP8266-01 having specified Baud rate. A lot of third party can therefore work on any system like windows, Mac, or Linux.
manufacturers produce different modules based on this chip. This entire model is linked to a local network and can be
managed from anywhere by anyone accessible in that specific
J. Power Supply local network. MJPG streamer conducts live streaming. We
A power supply is a component of the hardware which proposed a framework to create a live streaming and monitoring
provides power to an electrical device. It derives input from an system in real time using this whole model is linked to a local
electrical outlet and converts the current to DC (direct current) network so it can be operated from anywhere by anyone
from AC (alternating current), which is what the device needs. accessible in that specific local network. MJPG streamer
It also controls the voltage to a reasonable level, allowing the conducts live streaming. Raspberry pi, with Wi-Fi connectivity
device to operate smoothly without overheating. The power installed. The pi will record the video of the location in real-
supply is an integral part of any computer and must work time during monitoring process. Video capture is done via
properly for the rest of the components to operate Simply find commands that are provided to the raspberry pi through the
the input where the power cord is plugged in, you can locate the device.
power supply on a system unit. [4] Raspberry pi connection with the motor driver is done
using Raspberry Pi's General Purpose Input Output (GPIO)
3. Literature Review pins. The GPIO pins are connected to the motor-shield input
[1] Network video capture framework is introduced in this pins. The motor shield's exit pins are connected to the motors.
 International Journal of Research in Engineering, Science and Management 435
Volume-3, Issue-8, August-2020
journals.resaim.com/ijresm | ISSN (Online): 2581-5792 | RESAIM Publishing

The motor shield and the raspberry pi are connected by a quickly while spending more calculation on promising regions
portable charger with 2-amp current. Upon proper attachment like objects Might be the cascade Seen as a special focus-of-
the raspberry pi is able to boot up. For controlling the motors, a attention mechanism for object Which, unlike previous
Python program is written where the GPIO pins will deliver the approaches, provides statistical guarantors who are unlikely to
output from the raspberry pi to the shield. The robot movement have the object of interest in discarded regions, In the face
is controlled through the directions mentioned on the web page recognition domain the machine Returns equivalent detection
created using Hyper Text Markup Language (HTML) code and rates to the best previous systems. The Detector is used in real-
webpage Universal Resource Locator (URL) address. The time applications 15 frames per second without the use of
Raspberry Pi model B communicates this process through Wi- differentiating pictures or detecting skin colours.
Fi. The camera module is installed in its port and is enabled in [8] Embedded remote video control device, focused on ARM
raspberry pi configurations. MJPEG streamer is installed and and quick motion Stimulus algorithm (2011 International
configured for Live streaming of videos. Conference on Instrumentation, Measurement and Control)
[5] The Raspberry pi and Open CV Smart Security Camera Jian Wang Automation department, Harbin Automation
is a surveillance device built to be used within a warehouse Institute University of Technology 150080, Harbin
facility. Using a raspberry pi, this system is conceived using a Herlongjiang: Current H.264 codec-based video chaotic
low cost safety camera with night vision capability. This system encryption schemes may be applied to two classes: before
has the ability to detect and detect human smoke which can be H.264 encoding, the original video data is encrypted with
used to prevent possible crimes and potential fire. The chaos; after H.264 encoding, it is encrypted with chaos. The
researchers have developed a light-footed surveillance camera main disadvantage of the two classes of schemes is that the
that can identify the condition of the scene being monitored and contradictory problem between desirable safety and rapid frame
also notify or alert the event. This system also offers nighttime rate is not well resolved. This paper presents a novel H.264
security, as it has the potential to provide night vision. Night codec-based video-chaotic encryption scheme to cope with the
vision capability is accomplished by simply eliminating problem, where the original video data is encrypted by a stream
infrared (IR) filter from an ordinary webcam and can thus be cipher and position scrambling with chaos after H.264
used with the aid of the IR Light Emitting Diode illuminator for encoding. In particular, H.264 hardware encoding, multi-core
night vision sensing. The system can also use background multi-threading, H.264 data format protection and adaptive
subtraction algorithm to detect movement of an object. When a memory selection strategy are adopted for making the
moving body is detected, it may be marked as person or smoke suggestion.
by the system. When smoke is detected the device may alert fire
or unauthorized individual in the form of an alarm and email. 4. Flowchart
[6] Face recognition is still a very difficult task when the
A. Flowchart for Creating Database
image of the input face is noisy, obstructed by other obstacles.
Low-resolution, camera not facing and not properly lit. These The flowchart below shows the steps involved in creating an
problems contribute to the extraction of the feature and Face authorized user database:
recognition system unstable as a consequence. In this paper the
system proposed incorporates the novel idea of using hair-like
characteristics, which were widely used in Item recognition and
the probabilistic facial classification Reconnaissance. The
system proposed is simple, real-time, efficient and robust
against most of the problems described. Experimental findings
on public databases show the proposed. The system actually
outperforms state of the art facial recognition systems.
[7] The paper describes a machine learning approach for the
detection of visual objects which can process images Extremely
fast and attain high detection rates. That's it. Three key
contributions distinguish work The former Is the
implementation of a new representation of pictures, called
"Integral lineage" which enables our detector to measure its
features very quickly The second is a learning process Fig. 2. Flowchart for Creating Database
AdaBoost based algorithm, which selects a small number of
essential visual features and yields from a larger collection When the robot receives the power supply the pi camera
Extremely elliptical The third contribution consists of a method starts capturing the user's face in front of it. It captures the set
for combining ever more complex classier in a "cascade" that of images and stores them during face detection as a user
allows background regions of the classier Image to be discarded reference database for authentication. That authorized user
 International Journal of Research in Engineering, Science and Management 436
Volume-3, Issue-8, August-2020
journals.resaim.com/ijresm | ISSN (Online): 2581-5792 | RESAIM Publishing

database will consist of 30-50 images in order to obtain a high identify the image that marks the image pixels by
precision during face detection and to avoid detection of thresholding each pixel's neighborhood.
unauthorized face as allowed. The authorized person's created  A new binary value is set for each neighbor of the
image database is converted into matrix format for more rapid central value. For values equal to or higher than
and accurate detection. threshold, we set 1.
B. Flowchart for face detection and authentication  The result is converted as a binary number by taking
the clockwise direction of the threshold values.
The flow-chart below shows the entire Security Patrolling
 The binary is converted to decimal form and then
Robot operation:
converted to histogram, which is concatenated into a
larger histogram.
 The histogram obtained is compared to the Euclidean
distance reference histogram.
 Distance is known as the measure of trust.
 When the value of the trust is smaller than the
threshold, the other party is an intruder.

Fig. 4. Computation of LBPH (Local Binary Pattern Histogram)

6. Conclusion
This paper presented an overview on night vision security
patrolling robot using Raspberry Pi.

References
Fig. 3. Flowchart for face detection and authentication [1] M. L. V. N. B. S. Kumar, N. S. Murthi Sarma, Ch. Sridevi, A. Pravin
"ARM9 Based Real Time Embedded Network Video Capture and SMS
The camera detects the face in front of it when the robot is Alerting system."
[2] Krishna Kishore, B. Chinna Rao, P. M. Francis, "ARM Based Mobile
initialized and cross-checks it with reference database. The Phone- Embedded Real Time Remote Video Surveillance System with
captured face is sent to the training sequence where it is Network Camera,” International Journal of Emerging Technology and
translated to matrix format and is compared to the current Advanced Engineering, vol. 2.
database of references. If the face is authenticated, then the [3] Kavitha Mamindla, V. Padmaja, Ch. Naga Deepa, "Embedded Real Time
Video Monitoring System using Arm," IOSR Journal of Engineering, vol.
robot starts to move forward and to check continuously. If the 3.
face is not authenticated, then the sounding of the buzzer alerts [4] Tasleem Mandrupkar Manisha Kumari Rupali Mane, "Smart Video
the owner. Security Surveillance with Mobile Remote Control."
[5] Nava Jeevan Raju O. and Praveen P, "Video monitoring and motion
detection system based on arm-linux platform and http protocol with SMS
5. Face Detection and Recognition capability."
[6] Wang Liwei Yan Shi Xu Yiqiu, "A Wireless Video Surveillance System
Opencv has three built-in face recognizers: based on 3G Network," Conference on Environmental Science and
1. Eigen Faces Information Application Technology.
2. Fisher Faces [7] Zhaomin Zhu, Takashi Morimoto, Hidekazu Adachi, Osamu Kiriyama,
Tetsushi Koide and Hans Juergen Mattausch, “Multi-view Face Detection
3. Local Binary Patterns Histograms (LBPH) and Recognition using Haar-like Features.”
In the designed system we use Local Binary Patterns [8] P. Viola and M. Jones, “Rapid object detection using a boosted cascade
Histograms (LBPH). of simple features,” in Proc. of IEEE Conference on Computer Vision and
Pattern Recognition, Kauai, HI.
A. Local binary patterns histograms
 It is the duty of the face recognition algorithm to