Iris Recognition Repository / files

     1.1 --- a/reports/final/content/finalproduct.tex	Fri Mar 19 09:09:52 2010 +0000
     1.2 +++ b/reports/final/content/finalproduct.tex	Fri Mar 19 09:11:02 2010 +0000
     1.3 @@ -1,3 +1,40 @@
     1.4 +
     1.5 +\chapter{Final Results}
     1.6 +
     1.7 +With the batch command line option of the application, it is possible to
     1.8 +process a large amount of images of eyes. Using this option, 108 images of eyes
     1.9 +from the CASIA database were loaded into the application; the pupil, iris and eyelids were auto-detected to generate a bitcode and store it in the database. These were then compared to an alternative image of each eye 
    1.10 +to check for matches.
    1.11 +
    1.12 +
    1.13 +To further test primarily for false matches, the first three images of each eye were
    1.14 +loaded into the database, and then compared to a fourth
    1.15 +image for each eye. As there were originally three images for each eye, and
    1.16 +then a comparison was performed on all of the extra 108 images, roughly 35,000
    1.17 +iris comparisons took place ($(108 * 3) * 108$). 
    1.18 +
    1.19 +Our results of these tests are as follows: \begin{itemize} \item 0 false
    1.20 +		matches in $\sim$35,000 comparisons \item 70\% (75 out of 108) match rate
    1.21 +		with Professor Daugman's suggested 0.32 Hamming distance.
    1.22 +\end{itemize}
    1.23 +
    1.24 +The complete lack of any false matches and the incredibly high match rate are a
    1.25 +testament to the robustness and feasibility of an iris recognition system. The
    1.26 +tests have suggested that in virtually any case where the iris boundary is
    1.27 +correctly located, the iris should be correctly identified.
    1.28 +
    1.29 +As a roughly 70\% match rate was achieved for the iris location, this percentage has, as
    1.30 +predicted, been reflected in the overall match rate in the
    1.31 +database of images. Similarly to Daugman, we can also observe that the Hamming
    1.32 +distances produced by comparing different irides tends to follow a binomial distribution, with a mean around 0.46; see
    1.33 +figure \ref{hd}.
    1.34 +
    1.35 +\begin{figure}
    1.36 +  \centering
    1.37 +    \includegraphics[width=0.55\textwidth]{hd}
    1.38 +    \caption{The distribution of Hamming distances for one run of all the irides in the database}
    1.39 +    \label{hd}
    1.40 +\end{figure}
    1.41  \chapter{Conclusions}
    1.42  \section{Final Product}
    1.43  The final prototype is a functional iris detection application, which provides the 
    1.44 @@ -49,43 +86,6 @@
    1.45  thickness of the top eyelid, it can be seen that, although not perfect, a very
    1.46  reasonable estimate of the eyelid boundary is found automatically.
    1.47  
    1.48 -\chapter{Final Results}
    1.49 -
    1.50 -With the batch command line option of the application, it is possible to
    1.51 -process a large amount of images of eyes. Using this option, 108 images of eyes
    1.52 -from the CASIA database were loaded into the application; the pupil, iris and eyelids were auto-detected to generate a bitcode and store it in the database. These were then compared to an alternative image of each eye 
    1.53 -to check for matches.
    1.54 -
    1.55 -
    1.56 -To further test primarily for false matches, the first three images of each eye were
    1.57 -loaded into the database, and then compared to a fourth
    1.58 -image for each eye. As there were originally three images for each eye, and
    1.59 -then a comparison was performed on all of the extra 108 images, roughly 35,000
    1.60 -iris comparisons took place ($(108 * 3) * 108$). 
    1.61 -
    1.62 -Our results of these tests are as follows: \begin{itemize} \item 0 false
    1.63 -		matches in $\sim$35,000 comparisons \item 70\% (75 out of 108) match rate
    1.64 -		with Professor Daugman's suggested 0.32 Hamming distance.
    1.65 -\end{itemize}
    1.66 -
    1.67 -The complete lack of any false matches and the incredibly high match rate are a
    1.68 -testament to the robustness and feasibility of an iris recognition system. The
    1.69 -tests have suggested that in virtually any case where the iris boundary is
    1.70 -correctly located, the iris should be correctly identified.
    1.71 -
    1.72 -As a roughly 70\% match rate was achieved for the iris location, this percentage has, as
    1.73 -predicted, been reflected in the overall match rate in the
    1.74 -database of images. Similarly to Daugman, we can also observe that the Hamming
    1.75 -distances produced by comparing different irides tends to follow a binomial distribution, with a mean around 0.46; see
    1.76 -figure \ref{hd}.
    1.77 -
    1.78 -\begin{figure}
    1.79 -  \centering
    1.80 -    \includegraphics[width=0.55\textwidth]{hd}
    1.81 -    \caption{The distribution of Hamming distances for one run of all the irides in the database}
    1.82 -    \label{hd}
    1.83 -\end{figure}
    1.84 -
    1.85  \newpage
    1.86  
    1.87  \section{Source Code}