Sei sulla pagina 1di 6

CHAPTER I

THE PROBLEM AND ITS BACKGROUND


1.1 Introduction
Object detection in images has long been an active research in the field of
Computer and Machine Vision. With the advent of fast-paced trends in computer
and machine learning, real-time object detection and recognition became
possible
[1]
. Object detection is now applied to real-time footages and is essential
in information gathering for many research projects and different industrial,
commercial and security applications that lead to the development of
sophisticated systems and major advancement in computer vision. The vast
application of object detection imposes different requisites and constraints that
can lead to low success rate of previous object detection systems available.
Problems mainly arises in environments wherein there is less control and objects
are placed in arbitrary poses in clustered or blocked environment.
To meet specific requirements, many algorithms have been proposed and
developed through the years; implementation of mathematical and statistical
principles like the Gaussian and Laplacian derivatives, as well as the Haar
Transform are some of the bases in developing the learning algorithms
[2]
. It
enabled the existing algorithms to perform different processes that help the
system in the detection process. One of the most well-known systems was
Speeded Up Robust Features developed by Herbert Bay et al..
[3]
.They focused
their algorithm on the problem of object recognition or 3D reconstruction. Another
various system was Computer Vision System Toolbox which provides algorithms,
functions, and applications for the design and simulation of computer vision and
video processing systems.
[4]
It can perform object detection and tracking, feature
detection and extraction, feature matching and motion detection tasks.
The face detection system has received growing attention. Nevertheless,
detection and recognition of other objects such as vehicles are also highly
essential especially in its implementation in using unmanned aerial vehicles.
Autonomous mobile systems are used in place of ground vehicles and humans
when there is limited access to desired locations due to terrain and presence of
other obstacles
[5]
.
Unmanned aerial vehicles with cameras are new innovations used in
research projects and experiments that contribute to future development of the
field of object detection and recognition. Innovations using unmanned aerial
vehicles incorporating object detection systems are increasingly used in many
applications such as surveillance and environmental disaster monitoring
[5]
.
Development of this application is a promising project that will be of great help
not just for large companies but for improving the quality of life.

1.2 Objectives of the Study:
The primary objective of this study is to design and develop real time object
detection for video surveillance. For deeper analysis of the study, the following
objectives are set to serve as guide in the development of the system:
1. To determine preliminary requirements for the development of a system
that detects objects in real time.
2. To design and develop object detection software that can specifically
detect objects like balls, humans, and fire.
3. To implement object detection algorithm that can specifically detect
objects like balls, humans, and fire.
4. To conduct preliminary and final testing of the system considering the
specific parameters such as image features, sensitivity, lighting conditions,
and processing time.
5. To determine the maintenance requirement needed of the system.

1.3 Significance of the Study
This study entitled Real Time Object Detection for Video Surveillance aims
to develop a system that will detect object in real time. This is an effort to
accumulate new ideas and innovations in computer vision specifically in object
detection that can be mounted on unmanned aerial vehicles.
With the availability of this study, additional information in object detection can
be easily obtained for further research regarding cascade object detector detect
surf features. This will be an essential contribution to the field of Artificial
Intelligence and Object Recognition which are branches under the vast field of
Computer Vision. The study is also beneficial to the following:
Batangas State University. Success of this research will provide not just
additional information and knowledge to the students who would like to conduct
future research about object recognition and detection; it will also provide
additional security for the faculty, students and staff.
Proponents, the proposed design will help them discover new ideas,
enhance their knowledge about their field, and will improve them to use the ideas
they learned inside the classroom.
Engineering students. This study will enable them to enhance and widen
their knowledge in developing algorithms and the whole research process. This
study will help them be more familiar to the environment of different algorithm
and its data acquisition techniques and also guide them in achieving student
outcomes which are (b) ability to design experiments as well as to analyse and
interpret data, (c) ability to design a system, component, or process to meet
environmental, social, political, ethical, health and safety, manufacturability, and
sustainability in accordance with standards, and (k) ability to use techniques ,
skills and modern engineering tools necessary for engineering practice.
Future researchers. The study will serve as reference for future researches
in data acquisition dealing with object detection and recognition.

1.4 Conceptual Framework
This research project will present a conceptual framework that presents the
overall process of the research. The conceptual paradigm will show the input,
process and output flowchart that will serve as a guide in the course of study all
throughout its development to final stage. The design of the system considers the
system components, material specifications and maintenance. The input of this
research project comprises knowledge on real time object detection for video
surveillance, algorithm, compute vision system toolbox cascade object detector,
Surf image processing, Matlab, database, AutoFlight 0.1 Alpha and software
programming. Collecting information concerning with the topic will help in
constituting ideas considering the current study and will give out as a guide for
the design proper. The process of this design, as shown in Figure 1.1, contains
the design and layout, construction and testing and evaluation.
To be able to determine the performance and effectiveness of the system,
testing will be formulated. The removal of false positives
[16]
is also an important
step in ensuring the accuracy of the system. Finally, the output of this research
project will be a real time object detection video surveillance system.
















Figure 1.1 Conceptual Paradigm of the Study




INPUT
Knowledge Requirements
Algorithm
Cascade Object
Surf
Image Processing
Matlab
Database
AutoFlight 0.1 Alpha
Related Studies
Hardware Requirements
Computer
Camera

PROCESS
Data Gathering
Design and Layout
Algorithm
Computer Vision
System Toolbox
Graphical User
Interface
Removal of False
Positives
Hardware and
software interfacing
Assembly of the
Design
Testing and Evaluation
System Functionality
Trouble Shooting
OUTPUT
Real Time Object Detection
System
1.5 Scope and Delimitations
The scope of this study is focused on design and development of real time
object detection. The Cascaded Object and SURF algorithm is be used. This
system is a visual surveillance system that detects object in real time in indoor
and outdoor environments. The objects that will be detected and recognized will
be limited to ball, humans, and fire. Also, some specific parameters such as
image features, sensitivity, lighting conditions, and processing time will be
limited.
As part of the collaborative project of the Electronics Engineering and
Mechatronics Engineering program, the camera that will be used in this project
will be attached to an unmanned aerial vehicle or any web camera. The camera
of the AR Drone 2.0 has the following specifications; 90 degree wide-angle
diagonal lens camera, has CMOS sensor, has a video frequency of 60 frames
per second and has a resolution of 320x240 pixels. The algorithm and the
software that will be developed will not be limited to the use of the unmanned
aerial vehicle; it will also be usable in other systems like CCTV.

1.6 Definition of Terms:
Object Recognition - Object recognition is a process for identifying a
specific object in a digital image or video
[6]
.
Computer Vision a robot analogue of human vision in which
information about the environment is received by one or more video cameras and
processed by computer: used in navigation by robots, in the control of automated
production lines, etc.
[7]
Image Processing The study of any algorithm that takes an image as
input and returns an image as output.
[8]

Algorithm A step-by-step procedure for solving a problem or
accomplishing some end especially by a computer.
[9]
SURF (Speeded Up Robust Features) A robust local feature detector
that can be used in computer vision tasks like object recognition or 3D
reconstruction.
[10]

Computer Vision System Toolbox cascade object detector It can
detect object categories whose aspect ratio does not vary significantly. Objects
whose aspect ratio remains approximately fixed include faces, stop signs, or cars
viewed from one side.
[11]
Database A collection of files kept together within a computer
application that can be easily accessed and retrieved.
[12]

GUI A program interface that takes advantage of the computer's
graphics capabilities to make the program easier to use.
[13]

MATLAB a high-level language and interactive environment for
numerical computation, visualization, and programming.
[14]

Camera A camera is a device that is used to transfer an image onto
either a sheet of film or a space on a digital hard drive.
[15]

False Positives The erroneous identification of a threat or dangerous
condition that turns out to be harmless. False positives often occur in intrusion
detection systems.
[16]
Real Time Process of information that produces a response within a
specified time.
[17]

Autoflight An application used to control the flight of an AR Drone 2.0.
[18]

Surveillance The monitoring of the behaviour, activities, or other
changing information, usually of people for the purpose of influencing, managing,
directing, or protecting them.
[19]

Resolution Determined by the size of the units of information
representing an image.
[20]

Potrebbero piacerti anche