%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%2345678901234567890123456789012345678901234567890123456789012345678901234567890
% 1 2 3 4 5 6 7 8

\documentclass[letterpaper, 12 pt, conference]{ieeeconf} % Comment this line out

% if you need a4paper
%\documentclass[a4paper, 10pt, conference]{ieeeconf} % Use this line for a4
% paper

\IEEEoverridecommandlockouts % This command is only

% needed if you want to
% use the \thanks command
\overrideIEEEmargins
% See the \addtolength command later in the file to balance the column lengths
% on the last page of the document

% The following packages can be found on http:\\www.ctan.org

\usepackage{graphicx} % for pdf, bitmapped graphics files
\usepackage{bm}
\usepackage{amsmath,amssymb}
\usepackage{ifpdf}
%\usepackage{cite}
\usepackage{algorithmic}
\usepackage{array}
\usepackage{mdwmath}
\usepackage{pdfpages}
\usepackage{mdwtab}
\usepackage{eqparbox}
\newcommand{\uvec}[1]{\boldsymbol{\hat{\textbf{#1}}}}
%\usepackage{epsfig} % for postscript graphics files
%\usepackage{mathptmx} % assumes new font selection scheme installed
%\usepackage{times} % assumes new font selection scheme installed
%\usepackage{amsmath} % assumes amsmath package installed
%\usepackage{amssymb} % assumes amsmath package installed

\title{\LARGE \bf
Face Recognition Challenge using SIFT Algorithm
}

%\author{ \parbox{3 in}{\centering Huibert Kwakernaak*

% \thanks{*Use the $\backslash$thanks command to put information here}\\
% Faculty of Electrical Engineering, Mathematics and Computer Science\\
% University of Twente\\
% 7500 AE Enschede, The Netherlands\\
% {\tt\small h.kwakernaak@autsubmit.com}}
% \hspace*{ 0.5 in}
% \parbox{3 in}{ \centering Pradeep Misra**
% \thanks{**The footnote marks may be inserted manually}\\
% Department of Electrical Engineering \\
% Wright State University\\
% Dayton, OH 45435, USA\\
% {\tt\small pmisra@cs.wright.edu}}
%}

\author{Bhumiti Gohel $^{1}$ and Supan Shah$^{2}$% <-this % stops a space

\thanks{}% <-this % stops a space
\thanks{$^{1}$Bhumiti Gohel,AU1841051, Department of Information and Communication
Technology}%
\thanks{$^{2}$Supan Shah, AU2044011, Department of Computer science and engineering
with specialization in Data Science}%
}

\begin{document}

\maketitle
\thispagestyle{empty}
\pagestyle{empty}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{abstract}
The field of Computer Vision consists of many advanced techniques and edge
detection is the one of the fundamental bases for those techniques. Edges play an
important role in helping for doing analysis and interpretation of any Image in the
domain of computer image analysis. Robinson edge detection method is one of the
simplest methods to quickly detect edges using predefined filters.
\end{abstract}

\begin{IEEEkeywords}
\textbf{Keywords}-- SIFT algoritm, Computer Vision
\end{IEEEkeywords}

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\section{INTRODUCTION}
Edges play an important role in helping for doing analysis and interpretation of
any Image in the domain of computer image analysis. Edge detection is an image
processing approach for detecting the boundaries of objects within images.
Furthermore, There are three types of edges present in the image.i)Vertical edge,
ii)Horizontal Edge & iii)Diagonal edges. Edge detection performs a very crucial
role in image segmentation, pattern recognition, image morphology, and feature
extraction in the field of image processing, computer vision, and machine vision.
It can identify discontinuities in an image. Edge detection can be performed by
using various algorithms like Sobel operator, Canny edge detection, Prewitt
operators, Roberts and fuzzy logic methods which are very frequently used methods.
Moreover, some masking methods are also used for edge detection which are very
primitive methods like Robinson compass mask, Krisch compass mask, etc. Among all
of these, in this paper, we are going to use the Robinson Compass Masks method for
edge detection. We have built this algorithm from scratch. This method is also
known as direction mask because it detects eight major compass orientations and
each of them will detect the edges with respect to its direction. Therefore, in
this method, we take one mask and rotate it in all 8 directions so that we can
detect the edges from all different angles.

\section{MOTIVATION}
Edge Detection process comprises of extracting edge line with orientation same as
original image without changing the properties of original image. Major discrete
values of image are brightness and contrast as mentioned in [1]. They generally
corresponds to:
\begin{itemize}
\item Depth Discontinuities
\item Surface orientation Discontinuities
\item Changes in material properties
\item Variations in scene brightness
\end{itemize}

\section{BACKGROUND}
Historically, Edge detection has been used to detect low-level feature detection.
There are several algorithms for edge detection. The Robinson compass mask is one
of the simplest and the most primitive method among all of them. Robinson Detection
method calculates an approximation of the first derivation of the image data and
uses it as an edge detector. We take one mask and rotate it in all 8 directions. As
a result, we can detect the edges from all different angles. In the following image
8 types of directional mask are depicted.
\begin{figure}[h]
\centering
\includegraphics{Robinson-kernels-and-their-orientation.png}
\caption{Robinson compass masks or edge detection masks}
\end{figure}
As we can observe here, all the masks contains only the values 0, 1, -1, 2, -2.
Also, these masks are symmetrical. Therefore, we have to calculate only 4 masks and
the other 4 marks are the negation of the first result. The convolution of the
masks with an image will create a high-value output where there is a rapid change
in pixel value in original image[3]. This way we can detect an edge.

\section{DETAILED MATHEMATICAL ANALYSIS OF ROBINSON COMPASS MASK ALGORITHM}

Robinson Detection method calculates an approximation of the first derivation of
the image data and uses it as an edge detector. We take one mask and rotate it in
all 8 directions. As a result, we can detect the edges from all different angles.To
detect the edges,we use convolution operation between the gray-scale converted
image and the 3x3 filter.The operation is as follows:
\begin{figure}[h]
\includegraphics[width=7.5cm, height=3cm]{convo.jpg}
\caption{Convolution Operation}
\end{figure}
Output I' will be the output of the robinson compass. This is the Robinson Compass
Mask algorithm.
\\
\\
\section{EXPERIMENTS AND RESULTS}
In this paper, we have used an image dataset from the Kodak Corporation for
unrestricted research usage. In this dataset[2], there are total of 25 uncompressed
PNG true-color images of size 768 × 512 pixels. After collecting the dataset we
converted color images into grayscale for better analysis and visualization. After
that, these images were passed into the different types of 8 directions kernel.
More output results of this Algorithm performed on various types of images can be
found here[4].
\\
\\
\\
\\
\\
\\
\\
\\
\textbf{Results:}
\begin{figure}[h]
\centering
\includegraphics[width=5.5cm, height=5cm]{kodim07.png}
\caption{Original Image}
\end{figure}
\\
\\
\begin{figure}[h]
\centering
\includegraphics[width=8.5cm, height=10cm]{PicsArt_02-05-06.32.25.jpg}
\caption{Output: Robinson Compass Mask}
\end{figure}

\begin{figure}[h]
\centering
\includegraphics[width=5.5cm, height=5cm]{kodim15.png}
\caption{Original Image}
\end{figure}
\begin{figure}[h]
\centering
\includegraphics[width=8.5cm, height=10cm]{final2.jpg}
\caption{Output: Robinson Compass Mask}
\end{figure}

\section{Comparison}

\subsection{Comparison between our Robinson Compass Mask and In-built Function}

We get smoother output with inbuilt function as compared to our algorithm as we
have not used any special function for smoothing.
\\
\\
\subsection{Comparison between Robinson Compass Mask And Other Algorithm}
There are several algorithm for edge detection.We compare some of the algorithm
like krisch Opertor,Canny Edge detection,Prewitt Edge detection below.
\\
\\
\begin{figure}[h]
\centering
\includegraphics[width=10.5cm, height=8cm]{Table_compare.jpg}
\caption{Comparison of Edge Detection Algorithm}
\end{figure}
Below shows a concise comparison:
\\
\begin{figure}[h]
\centering
\includegraphics[width=6.5cm, height=7cm]{krish_4.png}
\caption{Comparison of Edge Detection Algorithm}
\end{figure}
\begin{figure}[h]
\centering
\includegraphics[width=7.5cm, height=7cm]{pri_4.png}
\caption{Comparison of Edge Detection Algorithm}
\end{figure}
\begin{figure}[h]
\centering
\includegraphics[width=8.5cm, height=5cm]{canny_4.png}
\caption{Comparison of Edge Detection Algorithm}
\end{figure}
\section{CONCLUSION}
The experiments conducted in the report show us how effective the Robinson compass
method is at highlighting the edges in a given image. This method gives clear
representations of the edges as well as provides flexibility of detecting edges in
a particular direction only. This simplicity and flexibility renders the Robinson
method as one of the top choices for low-level feature detection. The experiment
also observed the need of smoothing to better the results of filter based edge
detectors. In the end, the Robinson compass method provides a simple, quick as well
as flexible method for edge detection.

\\
\\
\\
\\
\begin{thebibliography}{99}

\bibitem{c1} https://www.ukessays.com/essays/biology/edge-detection-methods-in-
digital-image-processing-technology-essay.php
\bibitem{c2}http://www.cs.albany.edu/~xypan/research/snr/Kodak.html
\bibitem{c3}https://en.wikipedia.org/wiki/Robinson$_$compass$_$m
\bibitem{c4}https://drive.google.com/drive/folders/1p6rbF9Bk7EAMN-
HdtBa5cij30RSPOUye?usp=sharing

\end{thebibliography}

\end{document}