\documentclass[12pt,letterpaper]{article}
\pagestyle{plain}

\usepackage{geometry}
\geometry{papersize={216mm,279mm},total={176mm,239mm}}

\setlength{\parindent}{0pt}
\setlength{\parskip}{1em}

\begin{document}

\begin{raggedright}
  \hfill Keir Mierle (991656383)\\
  \hfill Aly Merchant (991579083)\\
\end{raggedright}

\vspace{2em}

\centerline{\LARGE \bf Condensation Project Proposal}
\vspace{1em}

This project will involve implementing the condensation tracking
algorithm.  The first part of our project will involve translating the
simple condensation code online into C++.  It is a very simple
implementation of condensation limited to 1 dimension.  This will give
us some overview of the algorithm and should not take very long.

For our implementation of condensation we will use the same basic
structure as our translated version.  The core of our project will be
the implementation of the observation and dynamic models.  We will first
improve the observation model by limiting the region of interest to a
curve encompassing the object.

The second part will be the creation of a dynamic model.  For the
initial testing we will be using a generic random walk model.  However,
we would like to use a second order model as outlined in the original
condensation paper.  The model parameters will be set to reasonable
defaults rather than be learned from the input.

Our project will be written in C++, we will reduce some of the
complexity by using outside libraries for video processing and some of
the more intensive math (in general if something can be done in matlab
we will find a library to handle it instead of reinventing it).  We are
currently considering  the boost libraries ({\tt http://www.boost.org/})
as well as the OpenCV library 
({\tt http://www.intel.com/research/mrl/research/opencv/}).  We will
also be using the simple condensation code 
({\tt http://www.robots.ox.ac.uk/$\sim$misard/download.html}) for reference.

We will be using video from online physics demos found at \\
{\tt http://www.doane.edu/Dept\_Pages/PHY/PhysicsVideoLibrary/videolibrary.html}.
The videos we will be using will involve some collisions and should be
able to demonstrate some of the improvements of condensation compared to
other solutions (such as Kalman filters).

As our references we will be using the original condensation paper as
well as Active Contours (a book available online at:
{\tt http://www.robots.ox.ac.uk/$\sim$contours/}).  The book was also written
by the same authors but goes into more detail, and we will be using it
to fill in where the paper becomes mathematically dense.

\end{document}