NeoGeographyToolkit/StereoPipeline

Subversion checkout URL

You can clone with
or
.
Fetching contributors…

Cannot retrieve contributors at this time

68 lines (48 sloc) 3.0 kB
 \title{LMMP: Stereo Module Testing Plan} \author{ Ara V. Nefian \and Michael Broxton } \date{\today} \documentclass[12pt]{article} \begin{document} \maketitle This document describes a testing plan for the LMMP stereo module. {\underline {\bf Accuracy measures}} \begin{itemize} \item {\bf Option 1:} Short term, if ground truth data (Apollo Panoramic images) is available (Feb 15). Let the predicted horizontal and vertical disparity map for the $k$th image be denoted as ${\bf P}_{k, h}(i,j)$ and ${\bf P}_{k, v}(i,j)$ respectively with the corresponding ground truth dispariy map ${\bf T}_{k, h}(i,j)$ and ${\bf T}_{k, v}(i,j)$ for all pixels $(i,j)$. The ${\bf average}$ ${\bf error}$ is ${\bf E} =\frac{1}{K}\sum_k\frac{\sum_{ij} (P_{k,v}(i,j)-T_{k,v}(i,j))^2 + (P_{k,h}(i,j)-T_{k,h}(i,j))^2}{N_{k,pred}}$ where $N_{k,pred}$ is the number of pixels for which the disparity map was computed in the $k$th image, and $K$ is the total number of images for which we have ground truth. The ${\bf average}$ ${\bf coverage}$ ${\bf C}$ measures the number of pixels for which the disparity is determined: ${\bf C} = \frac{1}{K}\sum_k\frac{N_{k,pred}}{N_{k,true}}$, where $N_{k, true}$ is the resolution of the ground truth images. The goal is to have a small average error ${\bf E}$ and large avearge coverage ${\bf C}$ measures. \item {\bf Option 2:} Very short term, before the ground truth data is made available (Jan 15th). Compute the ${\bf E}$ measure using existing synthetic ground truth data, and the ${\bf C}$ measure using the MOC and Apollo Metric data. \item {\bf Option 3:} On the longer term the ${\bf E}$ measure will be weighted by the confidence score of the Bayesian subpixel correlator. ${\bf \tilde E} =\frac{1}{K}\sum_k\frac{\sum_{ij} ((P_{k,v}(i,j)-T_{k,v}(i,j))^2 + (P_{k,h}(i,j)-T_{k,h}(i,j))^2)\frac{p_{ij}}{\sum_{ij}p_{ij}}}{N_{k,pred}}$ \end{itemize} {\underline {\bf Performance measures}} Very short term (Jan 15th). Compute the number of additions, multiplications log and exp operations per pixel and run time required to generate the dense disparity map. %\begin{abstract} %This is the paper's abstract \ldots %\end{abstract} %\section{Introduction} %This is time for all good men to come to the aid of their party! %\paragraph{Outline} %The remainder of this article is organized as follows. %Section~\ref{previous work} gives account of previous work. %Our new and exciting results are described in Section~\ref{results}. %Finally, Section~\ref{conclusions} gives the conclusions. %\section{Previous work}\label{previous work} %A much longer \LaTeXe{} example was written by Gil~\cite{Gil:02}. %\section{Results}\label{results} %In this section we describe the results. %\section{Conclusions}\label{conclusions} %We worked hard, and achieved very little. \bibliographystyle{abbrv} \bibliography{simple} \end{document} This is never printed
Something went wrong with that request. Please try again.