Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology

ISSN No:-2456 2165

Theft-Detection using Motion Sensing Camera

Ajay Kushwaha Ankita Mishra
Department of Information Technology Department of Information Technology
Atharva College of Engineering Atharva College of Engineering
Mumbai, India Mumbai, India
[email protected] [email protected]

Komal Kamble Rutuja Janbhare

Department of Information Technology Department of Information Technology
Atharva College of Engineering Atharva College of Engineering
Mumbai, India Mumbai, India
[email protected] [email protected]

Abstract Theft of valuable objects is some of the Also, output device must be able to be turned off using a
never-ending problems in the world. Several detecting keyboard or click on the mobile phone. For this project, the
devices are already available in the market today. system will be used to detect motion in the shop or any
Example of such is to provide an alarm which can be protected area using the camera.
activated using a remote control. Unfortunately, these
devices had no provision to automatically detect theft
attempts by alarming the owner. To solve this problem, II. EASE OF USE
Camera footage based theft detection will be used in this
project with the help of image processing to detect theft The system consists of simple hardware structure and hence
occurrence. The system will capture the image frames it would be easier for the user to arm and disarm the system
and by comparing the two frames, it will be able detect according to his/her needs. The owner will just have to install
motion and immediately an alert message will be sent to the camera at the ceiling of the room or the shop and provide
the owner along with captured image and options such power supply to it. Once the system is started, it will detect
as neglect, call the police or fire brigade. motion and send an alert message to the owner. Since an
application will also be built for the easy interaction of the
KeywordsAnti-Theft Device, Image processing, user with the system such as START and STOP.
Raspberry Pi, RANSAC, Camera module, Motion Detection
III. SYSTEM DESIGN
I. INTRODUCTION
To resolve the problems of existing work we propose a new
Theft is one of the most common and oldest criminal approach, which would be helpful to remove the unnecessary
behaviors and it is increasing day by day. The rate of crime use of hardware and the cost of the project will be reduced.
has increased not only in the single area in the world but also The proposed work will solve the problems of false alarm by
in worldwide. Due to increase rate of theft, the people have removing outliers using various algorithms such as frame
suffered in fear and loss. In order to stop this increasing theft differencing, segmentation, etc.
across the world, there is a need for a theft deterrent system
that is convenient in use, relatively free from false alarms and
does not require frequent user action to arm and disarm the
system.
An anti-theft system will be developed which will mainly
focus specifically on the room security system. This device
will be installed inside the room, mounted on the ceiling will
detect motion and it would be efficient enough to
differentiate between the false and real theft by detecting
motion and differentiating between the small and large
changes that are occurring in the frame.
Software/Application is used to activate the output device by
a click of a button or simply hit the key on the keyboard.

IJISRT17NV31 www.ijisrt.com 90
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

A. Block Diagram:

Figure 1: System Block Diagram

System Block Diagram i.e. figure 3.1 below served as the

framework of the system. The Raspberry Pi 3 and Raspberry
Pi Camera module is the main components of the system. It
serves as a controller of the entire process. Wi-Fi module is
built in the Raspberry Pi 3 b hardware, so there will be no
need to add Wi-Fi module externally. The proposed system
will be an intelligent system and it can eliminate the need for
continuous surveillance by human resource. Thus, any
human extra work is ruled out. In this project raspberry pi 3
B (model) will be used as heart system. The system will
continuously check the status of place by sensors that, Is
anyone entering the shop or not and sends the alert message
to the owner with the captured images. In this security
system motion will be detected by using image processing
techniques.

A. Flowchart:
The below flowchart i.e. figure 4.2 gives the complete flow
of working of the system from the capturing of an image to
sending the alert message to the owner.

Figure 2: Flowchart

IJISRT17NV31 www.ijisrt.com 91
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

1. Start: Camera Input

After installing the Raspberry Pi Camera module inside the
room or shop, the Anti-theft device will require a 5-12V
power supply. The whole system will start working on
providing the continuous power supply to the device.
2. Enter Theft
Once the system is installed and sufficient power supply is
provided to it, it will start capturing the image. After closing
the shop or room, if any motion happens in front of the
camera it will detect it as Theft entry and start following the
next steps of its algorithm.
3. Image Acquisition
The proposed system acquires the input video by using
Raspberry Pi camera module which will capture video at a
resolution of 280 180 pixels. The camera will be fixed on
the wall and capture video at a frame rate of 18 frames/sec. N corresponding matches are randomly selected and the rigid
To detect moving objects becomes quite complicated in the transformation Htt-1 is calculated. The objective function used
presence of noise, reflections, shadows, illumination in the RANSAC algorithm determines whether matches can
conditions. The proposed algorithm will be used to detect the be inliers by measuring the Euclidean distance between the
difference between the frames [7]. 3D points and the transformed 3D points. This means that a
4. Pre-processing 3D point transformed by Htt-1 should be located at the
matched 3D point. The most voted candidate is selected as
The pre-processing phase includes various interlinked the transformation matrix in the RANSAC algorithm [2].
activities such as background estimation, background
subtraction, Outlier rejection, Frame Differencing and d) Segmentation
Segmentation. The frame subtraction result contains many undesired points
a) Background Estimation such as ghost points and occlusion regions even though the
registration is accurate. Therefore, these points will be
Background estimation deals with estimating background of removed using segmentation technique. The segmentation is
an image. For the purpose of background estimation, a started at the outliers of the RANSAC whose points are on
reasonable amount of captured frames will be processed in the object after moving and its connected voxels are merged.
order to estimate the background. Background estimation If there is an occupied voxel in the 26-neighborhoods of the
approach deals with capturing the background only; this is given voxel, the occupied voxel will be said, connected to the
the simplest case for background estimation, as there is no given voxel. All connected voxels are selected using the 26-
other object in the image and the whole frame is considered neighborhoods until every neighbor of its neighbor has no
as the background. A single frame is sufficient for connected voxels. The process starts at voxel v and
background estimation in such case [7]. connected voxels are found [2].
b) Background Subtraction/Frame Differencing 5. Decision Making-I
Background estimation is followed by background The proposed system has a very simple decision-making
subtraction in every image processing algorithm. algorithm, based on feature matching. The first captured
Background subtraction is the most commonly used image will be stored in the database as the reference image.
technique for object detection. It deals with subtracting Object detected as a result of the new current image is
background from the image in order to detect object matched with the database values using the technique of
components from the image. Background subtraction is done feature matching. If there is a match between the current
in pixels domain, where the process is applied pixel by pixel. image frame and the reference image frame stored in the
In proposed system pixel by pixel, background subtraction is dataset, the proposed system will send an alert message as
done with a tolerance 35 [2]. well as the captured image as compared to the large original
c) Outlier Rejection video to the user/owner. If there will be no match between
the two frames, then it continues the process from the image
The nearest neighbor search will be used as the matching acquisition steps [7].
method between the previous and current images. However,
this method has drawbacks generating many outliers. These 6. Decision MakingII
outliers could lead to large registration error. The RANSAC The decision-making algorithm depends on the user/owner of
algorithm is used to remove outliers. The Below given the shop, after getting the alert and captured the image. The
algorithm shows the outlier rejection method with RANSAC. user will see the image and then decide whether the theft is
real or not and whether the action has to be taken or not.

IJISRT17NV31 www.ijisrt.com 92
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

7. Take Action B. Use-Case Diagram

After watching the image, if the user finds it real theft then The below Use-case diagram i.e. figure 3.3 describes the
he can call the police using the Call Police option provided overall role of the Anti-Theft device and the user.
in the application of the user or neglect it by clicking the
Neglect option.

Figure 3: Use Case Diagram

1. Device/System owner, the owner will check the image that will be sent along
with the message and decides whether it is a real theft or not.
The Anti-Theft device, once installed will be activated by the
If the theft is real, the owner will take action against that and
owner and then the motion detection work will be started by
click on the Call Police option on the application and ask
the camera with the help of written code for image
for help with police and if the message is a false alarm, then
processing techniques. If motion will be detected, then it will
the owner will neglect it by clicking on the Neglect option.
send the alert message to the owner along with the captured
image. The application on the owner side provides an option
to the user for handling the received alert message.
C. Data Flow Diagram
2. Owner/User
The DFD level 1 i.e. figures 3.4, shows the complete flow of
The device will require 5-12V of continuous power supply, the data and system from the camera that is capturing the
which should be provided by the owner before starting it. If image to the owner/user who will take action against the alert
the device detects motion and sends an alert message to the message. First of all, after starting the device, it will start

IJISRT17NV31 www.ijisrt.com 93
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

capturing the image frame using Raspberry Pi Camera difference will not present then it will end that process and
Module and then the noise errors/outliers will be removed start new comparisons of newly captured frames. The device
using RANSAC algorithm. The cleaned image frame as a will send an alert message and captured image frame at the
reference image and the next image frames as the current time of motion if any motion is detected. The owner will see
image will be stored in the Image Database. the message and image and decide what action to be taken
against it. If the user finds it as a real theft, then he/she can
After storing the image frames, it will compare both the
call the police or neglect it, using the options provided in the
reference and current frames and if it will find any difference
mobile application to control the device.
then it will conclude that motion is detected and if the

Figure 4: DFD Level-1

IJISRT17NV31 www.ijisrt.com 94
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

D. Sequence Diagram

Figure 5: Sequence Diagram

IJISRT17NV31 www.ijisrt.com 95
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

The Sequence diagram of Anti-Theft device i.e. figures 4.5 V. CONCLUSIONS

gives the clear idea of how the objects such as device, owner,
and police will work together in sequence. When the device The system will capture images only when the motions
will start, it will capture images and detect motion by exceed a certain threshold that is pre-set in the system. It thus
comparing the frames and if any theft will enter into the reduces the volume of data that needs to be reviewed. Also, it
shop/room, it will detect that action as motion and send an will help to save data space by not capturing static images
alert message to the owner along with the captured frame. which usually do not contain the object of interest. It will be
This whole process from the device end is combined into a applicable to both office and home uses. After successfully
single loop box because the above-mentioned process will implementing the project, it can be applied to the motion
run again and again in a loop. detection for a smart home security system which would be
very much helpful in auto theft detection for security
Once the alert message will be received by the owner on the
purpose. It can also be useful in bank, museum, and street at
application, the owner will have to decide whether the theft
midnight. As a future work, we can introduce Artificial-
is real or its just a simple movement and false alarm by
Intelligence into the system to distinguish between real theft
seeing the images sent by the device. Since, there are two
and just a simple movement of the object.
possible situations that theft could be real or false alarm,
hence alt box is used to represent both the alternatives. In
the first alternative, if condition (Theft: Real) is true then the
VI. ACKNOWLEDGMENT
owner will call the police and ask for help and in the second
alternative, the condition (Theft: False alarm) will be
checked and if it is true then the owner will neglect the alert It gives us great pleasure in presenting this paper titled:
message. Theft Detection using Motion Sensing Camera and we are
thankful to IJISRT (International Journal of Innovative
The sequence of all the roles of objects is shown in the
Science and Research Technology) for giving us the
sequence diagram of an anti-theft device and the loop
opportunity to research and publish this paper.
containing the process i.e. capture images, detect motion and
send an alert message to the owner will run in the loop for
We express our gratitude to our project guide Prof. Amruta
the continuous surveillance by the device and hence the
Pokhare, who provided us with all the guidance and
human effort to focus continuously will be reduced.
encouragement and helped us in finding our mistakes and
improving it. We also would like to deeply express our
sincere gratitude to Project Coordinators. We are eager and
IV. EXPECTED RESULTS
glad to express our gratitude to the Head of the Information
Technology Dept. Prof. Neelima Pathak, for her approval of
this project. We are also thankful to her for providing us the
With the approach to smart surveillance, the user will be
needed assistance, detailed suggestions and also an
allowed to depict a time for monitoring intruders. During the
encouragement to do the project.
survey time, the systems will continuously poll data from the
pi camera. In case of an intrusion, the motion will be
We would like to deeply express our sincere gratitude to our
detected by the Pi camera and sent it to raspberry pi for
respected principal Prof. Dr. Shrikant Kallurkar and the
further analysis. If the image processing flags a possible
management of Atharva College of Engineering for
intruder, the user is notified through the alert message along
providing such an ideal atmosphere to build up this project
with the captured image. The user can also view and
with a well-equipped library with all the utmost necessary
download the footage of the intruder to be used as evidence.
reference materials and up to date IT Laboratories.
Along with all these features, the overall results which are
expected from the system are as follows:
REFERENCES
An anti-theft device which can give a real-time alert
message to the owner. [1]. Rhowel Dellosa, Development of an Anti-Theft
A user-friendly application for the owner to control the Device using Motion Detection and Body
device i.e. when to start or stop. Temperature, Asia Pacific Journal of Multidisciplinary
Research, December 2014. P-ISSN 2350-7756 | E-ISSN
An efficient device that has the ability to distinguish 2350-8442.
between the real theft and false alarm. [2]. Yong-Deuk Shin, Jae-Han Park, Ga-Ram Jang, Moon
A device that can give authority to the user or owner to Hong Baeg, Moving Objects Detection using Freely
take actions against the alert message through options. Moving Depth Sensing Camera, 21st International
Conference on Pattern Recognition, November 2012.
[3]. Dixit Suraj Vasant, Babar Apeksha Arun, Meher Priya
Shivaji,Raspberry-Pi Based Anti-Theft Security
System with Image Feedback, Journal Of Information,
Knowledge, And Research in Electronics and

IJISRT17NV31 www.ijisrt.com 96
Volume 2, Issue 11, November 2017 International Journal of Innovative Science and Research Technology
ISSN No:-2456 2165

Communication Engineering, October 2017. ISSN:

0975 6779.
[4]. Rahane Madhuri, Pathak Mayuri, Wadekar Ashabai,
More Rupali, Study of Theft Detection and Tracking
using Raspberry Pi and PIR Sensor, International
Journal of Latest Trends in Engineering and
Technology (IJLTET), September 2015. ISSN: 2278-
621X.
[5]. Rahul Raguram1, Jan-Michael Frahm1, and Marc
Pollefeys, A Comparative Analysis of RANSAC
Techniques Leading to Adaptive Real-Time Random
Sample Consensus, ECCV 2008, Part II, LNCS 5303.
[6]. Ashwini Patil, Shobha Mondhe, Tejashri Ahire, Gayatri
Sonar, Auto-Theft Detection using Raspberry Pi and
Android App, International Journal of Research in
Engineering Application & Management (IJREAM),
October 2016. ISSN: 2494-9150.
[7]. Ali Javed and Sidra Noman, Theft Prevention Alarm
Based Video Summarization Algorithm, International
Journal of Information and Education Technology,
February 2012.
[8]. Mayur J. Charadva, Ramesh V. Sejpal, Dr. Nisha P.
Sarwade, A Study of Motion Detection Method for
Smart Home System, International Journal of
Innovative Research in Advanced Engineering
(IJIRAE), June 2014. ISSN: 2349-2163.

IJISRT17NV31 www.ijisrt.com 97