this is for holding javascript data
Volker Strobel edited chapter_Background_and_Related_Work__.tex
almost 8 years ago
Commit id: 436489065fb59922c155b87a101453ca06f619ce
deletions | additions
diff --git a/chapter_Background_and_Related_Work__.tex b/chapter_Background_and_Related_Work__.tex
index c5d0dd1..704f932 100644
--- a/chapter_Background_and_Related_Work__.tex
+++ b/chapter_Background_and_Related_Work__.tex
...
contrast them to the proposed method. Important
background knowledge for later chapters, concerning feature detection,
image matching, and textons will be introduced. While the range of methods for indoor localization is wide-spread---from laser range scanners over depth cameras to RFID tag based localization---only methods that are closely related to the proposed approach are discussed. In detail, this means that the approach should be based on a monocular camera.
\section{Vision-based Localization Methods}
\subsection{Fiducial Markers}
\label{sec:fiducialmarkers}
Fiducial markers, which are often used in augmented reality
applications~\cite{kato1999marker,garrido2014automatic}, have been
used for UAV localization and
landing~\cite{eberli2011vision,bebop2015}.
The markers encode information by the spatial arrangement of
black and white or colored image patches, and their corners can be
used for estimating the camera pose at a high frequency, using
triangulation methods. The positions of the markers in
an image are usually determined using local thresholding, which identifies salient image
regions. The positions are further refined by removing improbable shapes,
yielding an adjusted version of possible marker positions.
During flight, motion blur and varying distances can hinder the
detection of markers. Moreover, these markers might be considered as
visually unpleasant and might not fit into a product or environmental
design~\cite{chu2013halftone}. They offer little
flexibility, since one has to rely on predefined marker dictionaries.
There is no direct way to modify the time complexity of the algorithm.
% TODO: which errors have been achieved