Changeset 12281

Timestamp:

06/30/2009 07:48:06 PM (14 months ago)

Author:

dstn

Message:

review section on astrometric calibration

Location:

trunk/documents/theses/dstn

Files:

• figs-review/ngc3521-index.png (added)
• figs-review/ngc3521-sources.png (added)
• figs-review/ngc3521-zoom0-bw.png (added)
• figs-review/ngc3521-zoom1-bw.png (added)
• figs-review/ngc3521-zoom2-bw.png (added)
• figs-review/ngc3521.jpg (added)
• preamble.tex (modified) (1 diff)
• review.bib (modified) (1 diff)
• review.tex (modified) (10 diffs)
• thesis.bib (modified) (1 diff)
• thesis.tex (modified) (1 diff)

trunk/documents/theses/dstn/preamble.tex

@@ -54 +54 @@
 \newcommand{\Fig}{Figure\xspace}
 \newcommand{\fig}{figure\xspace}
+\newcommand{\figs}{figures\xspace}
 \newcommand{\figref}[1]{\xref{\fig}{#1}}
 \newcommand{\Figref}[1]{\xref{\Fig}{#1}}

trunk/documents/theses/dstn/review.bib

@@ -7 +7 @@
     volume = {65},
     pages = {43--72}
+}
+
+@article{barroncleaning,
+  author={Jonathan T. Barron and Christopher Stumm and David W. Hogg and Dustin Lang and Sam Roweis},
+  title={CLEANING THE USNO-B CATALOG THROUGH AUTOMATIC DETECTION OF OPTICAL ARTIFACTS},
+  journal={The Astronomical Journal},
+  volume={135},
+  number={1},
+  pages={414-422},
+  url={http://stacks.iop.org/1538-3881/135/414},
+  year={2008},
 }
 

trunk/documents/theses/dstn/review.tex

@@ -56 +56 @@
 reference frame is known as ``solving the astrometry'' or
 ``calibrating the astrometry'' of the image.  The \emph{blind
-  astrometry} task is to calibrate the astrometry of an image using
-only the image itself.  The broad goal of the \an project was to build
-a system that would allow us to create correct, standards-compliant
-astrometric \metadata for every useful astronomical image ever taken,
-past and future, in any state of archival disarray.  This is part of a
-larger effort to organize, annotate and make searchable all the
-world's astronomical information.
-
+astrometric calibration} task is to calibrate an image using only the
+image itself.  The broad goal of the \an project was to build a system
+that would allow us to create correct, standards-compliant astrometric
+\metadata for every useful astronomical image ever taken, past and
+future, in any state of archival disarray.  This is part of a larger
+effort to organize, annotate and make searchable all the world's
+astronomical information.
 
 
@@ -69 +68 @@
 scientists, the remainder of this chapter reviews the area of pattern
 recognition, and in particular the framework of \emph{geometric
-  hashing} for object recognition in images.  One of the contributions
+hashing} for object recognition in images.  One of the contributions
 of this thesis is to replace the simple hash table used in traditional
 geometric hashing with a \kdtree, so I review related work in hashing
 and other approaches for fast feature matching.  Finally, I present
-the blind astrometry task as an instance of object recognition, and
-review some previous approaches to the problem.
+the blind astrometric calibration task as an instance of object
+recognition, and review some previous approaches to the problem.
 
 %% Our approach?
 
 
-The remaining chapters present our approach to the blind astrometry
-problem.  \Chapref{chap:techreport} explains our approach and presents
-the results of large-scale tests of the system on real-world data.
-Chapters \ref{chap:verify} and \ref{chap:kdtree} delve into details of
-the approach: \Chapref{chap:verify} presents the Bayesian decision
-theory problem that lies at the heart of our approach, while
+The remaining chapters present our approach to the blind astrometric
+calibration.  \Chapref{chap:techreport} explains our approach and
+presents the results of large-scale tests of the system on real-world
+data.  Chapters \ref{chap:verify} and \ref{chap:kdtree} delve into
+details of the approach: \Chapref{chap:verify} presents the Bayesian
+decision theory problem that lies at the heart of our approach, while
 \chapref{chap:kdtree} explains the technical details of the \kdtree
 data structure implementation that is key to making our system
@@ -287 +286 @@
 % dot lamdan.dot -Tps2 -o lamdan.ps
 % ps2pdf -sPAPERSIZE=a4 lamdan.ps lamdan.pdf
-\includegraphics[height=0.9\textheight]{lamdan}
+\includegraphics[height=0.8\textheight]{lamdan}
 \end{center}
 \caption{Outline of the Geometric Hashing scheme.  This diagram is a
@@ -793 +792 @@
 have been proposed: in two surveys B\"ohm \cite{bohm2001} and
 Hjaltason \cite{hjaltason2003} identify B-, $\textrm{B}^{+}$-, ball-,
-bisector-, BSP-, DABS-, fq-, gh-, GNA-, hB-, $\textrm{hB}^{\pi}$-,
+bisector-, BSP-, DABS-, fq-, gh-, \mbox{GNA-,} hB-, $\textrm{hB}^{\pi}$-,
 hybrid-, IQ-, kd-, kd-B-, $\textrm{LSD}^h$-, M-, mb-,
-$\textrm{mb}^{\ast}$-, mvp-, oct-, post-office-, pyramid-, quad-, R-,
+$\textrm{mb}^{\ast}$-, mvp-, oct-, \mbox{post-office-,} pyramid-, quad-, R-,
 $\textrm{R}^{\ast}$-, $\textrm{R}^{+}$-, sa-, slim-, \mbox{sphere-,}
 SR-, SS-, TV-, vp-, $\textrm{vp}^{\ast}$-, and X-trees.  There is an
@@ -804 +803 @@
 
 
-\subsection{The astrometry problem}
+\section{Astrometric calibration as a pattern recognition task}
+
+
+\comment{
+wget "http://casjobs.sdss.org/ImgCutoutDR7/getjpeg.aspx?ra=166.45&dec=-0.03&scale=1&opt=&width=2000&height=2000" -O ngc3521-orig.jpg
+jpegtopnm ngc3521-orig.jpg | pnmrotate -45 | pnmcut 600 900 1400 1000 | pnmscale -reduce 2 | pnmtojpeg > ngc3521.jpg
+#---> http://live.astrometry.net/status.php?job=alpha-200906-68444159
+jpegtopnm ngc3521.jpg | ppmtopgm | pnminvert | pnmtojpeg > ngc3521-bw.jpg
+wget "http://live.astrometry.net/status.php?job=alpha-200906-36181848&get=field.xy.fits" -O ngc3521.xy
+wget "http://live.astrometry.net/status.php?job=alpha-200906-36181848&get=index.xy.fits" -O ngc3521-index.xy
+jpegtopnm ngc3521-bw.jpg | plotxy -N 100 -i ngc3521.xy -I - -x 1 -y 1 -C black -b white > ngc3521-sources.png
+jpegtopnm ngc3521-bw.jpg | plotxy -N 100 -i ngc3521.xy -I - -x 1 -y 1 -C black -b white -P | plotxy -I - -i ngc3521-index.xy -x 1 -y 1 -C black -b white -s crosshair -P | plot-constellations -f 18 -w ngc3521.wcs -i - -N -o ngc3521-index.png
+%%% wget "http://live.astrometry.net/status.php?job=alpha-200906-36181848&get=wcs.fits" -O ngc3521.wcs
+%%% scp gmaps:/data2/test-merc/tycho.mkdt.fits .
+wget "http://explore.astrometry.net/tile/get/?layers=tycho,grid,userboundary&arcsinh&wcsfn=alpha/200906/36181848/wcs.fits&gain=-0.5&bb=0,-85,360,85&dashbox=0.1&w=500&h=500&lw=3" -O ngc3521-zoom0.png
+wget "http://explore.astrometry.net/tile/get/?layers=tycho,grid,userboundary&arcsinh&wcsfn=alpha/200906/36181848/wcs.fits&gain=-1&bb=175.533,-17.7621663832,211.533,17.6598478619&dashbox=0.01&w=500&h=500&lw=3" -O ngc3521-zoom1.png
+wget "http://explore.astrometry.net/tile/get/?layers=tycho,grid,userboundary&arcsinh&wcsfn=alpha/200906/36181848/wcs.fits&gain=0.5&bb=191.733,-1.85338140354,195.333,1.74602498613&w=500&h=500&lw=3" -O ngc3521-zoom2.png
+for x in 0 1 2; do
+ pngtopnm ngc3521-zoom${x}.png | ppmtopgm | pnminvert | pnmtopng > ngc3521-zoom${x}-bw.png;
+done
+}
+
 
 % ~/an-2/usnob-map/execs/tilerender -x 0.000000 -y -85.000000 -X 360.000000 -Y 85.000000 -w 1024 -h 1024 -l 'tycho' -l 'grid' -l 'boundary' -s -g -0.5 -W 'tor/200706/51145570/wcs.fits' -L 5 -B 0.1 -d > tile1.png
@@ -820 +840 @@
 % http://oven.cosmo.fas.nyu.edu/test/status.php?job=tor-200706-51145570
 
+
 \begin{figure}
 \begin{center}
-\includegraphics[width=3.1in]{M65} \\
-\includegraphics[width=3.1in]{M65-sources} \\
-\includegraphics[width=3.1in]{M65-solved}
+\framebox{\includegraphics[width=0.99\figunit]{ngc3521-bw}} \\
+\framebox{\includegraphics[width=0.99\figunit]{ngc3521-sources}} \\
+\framebox{\includegraphics[width=0.99\figunit]{ngc3521-index}}
 \end{center}
-\caption{Top: input image (Copyright Volker Wendel, \texttt{http://www.spiegelteam.de/}).
-Middle: sources extracted from the image.
-Bottom: reference sources, transformed into the image coordinate system (green squares).
-Observe that while many of the image and reference sources are aligned, there are
-many image sources without reference sources, and at least one reference source without
-an image source.}
-\label{redgreen}
+\caption{\captionpart{Top:} Input image (credit: Sloan Digital Sky
+Survey).  \captionpart{Middle:} The brightest 100 sources extracted
+from the image.  \captionpart{Bottom:} Reference sources, transformed
+into the image coordinate system (crosshairs).  Many of the image and
+reference sources are aligned, but there are many image sources
+without reference sources.  Our system knows about the positions of
+many objects of interest on the sky, and has labelled the galaxy NGC
+3521.\label{fig:redgreen}}
 \end{figure}
+
 
 \begin{figure}
 \begin{center}
 \begin{tabular}{c@{\hspace{1pt}}c@{\hspace{1pt}}c}
-\includegraphics[width=1.6in]{M65-tile1c} &
-\includegraphics[width=1.6in]{M65-tile2c} &
-\includegraphics[width=1.6in]{M65-tile3c}
+\framebox{\includegraphics[width=0.31\textwidth]{ngc3521-zoom0-bw}} &
+\framebox{\includegraphics[width=0.31\textwidth]{ngc3521-zoom1-bw}} &
+\framebox{\includegraphics[width=0.31\textwidth]{ngc3521-zoom2-bw}}
 \end{tabular}
 \end{center}
-\caption{The location of the input image on the sky.  Left: the whole sky, in Mercator projection.  The dashed box
-  shows the zoomed-in region.  Middle: zoomed in by a factor of 10.  Right: zoomed in by a factor of 100; the
-  box shows the outline of the input image.}
-\label{onthesky}
+\caption{The location of the input image on the sky.
+  \captionpart{Left:} The whole sky, in Mercator projection.  The
+  sinusoid-shaped feature is the Milky Way.  The dashed box shows the
+  zoomed-in region.  \captionpart{Middle:} Zoomed in by a factor of
+  $10$.  \captionpart{Right:} Zoomed in by a factor of $100$.  The box
+  shows the outline of the input image.\label{fig:onthesky}}
 \end{figure}
 
 
-\emph{Astrometry} refers to the measurement of the positions and
-motions of celestial bodies.
-
-For modern astronomers, astrometry is often one of the first steps
-toward getting useful information out of an image of the sky.
-Aligning a new image with an \emph{astrometric reference catalog}
-(``solving the astrometry'' of the image) allows the astronomer to
-place the image within a standard coordinate frame.  This allows
-stars, galaxies, and other objects (\emph{sources}) in the new image
-to be identified with known sources (which in turn allows astronomers
-to calibrate other properties of the new image), and allows the
-positions of new sources to be described in a meaningful way.
-
-
-Several astrometric reference catalogs exist: one of the largest is
-the USNO-B1.0 catalog, created by the United States Navy Observatory,
-which contains position, motion, and brightness information for over
-one billion objects \cite{usnob,nomad}.
-
-\emph{Blind astrometry} describes the problem of solving the
-astrometry of an image given only the image itself.
-
-% This is equivalent to determining which stars are contained in the image.
-
-As part of the \an project, we are attempting to solve the blind
-astrometry problem for ``every useful astronomical image ever taken,
-past and future, in any state of archival disarray''\cite{an}.  As
+For modern astronomers, astrometric calibration is often one of the
+first steps toward getting useful information out of an image of the
+sky.  Aligning a new image with an \emph{astrometric reference
+catalog} allows the astronomer to place the image within a standard
+coordinate frame.  This allows stars, galaxies, and other objects
+(\emph{sources}) in the new image to be identified with known sources,
+which in turn allows astronomers to calibrate other properties of the
+new image, and allows the positions of new sources to be described in
+a standard reference frame.
+
+
+The task of blind astrometric calibration---automatically finding the
+astrometric calibration of an image, using only the information in the
+image pixels---can be seen as a pattern recognition problem.  As
 Bertin \cite{bertin2005} notes, ``astrometric and photometric
 calibrations have remained the most tiresome step in the reduction of
 large imaging surveys,'' so this is not only an interesting problem to
-solve, but one with practical implications for astronomers.  Figures
-\ref{redgreen} and \ref{onthesky} show sample results.  Given an input
-image, we do some image processing to find sources such as stars and
-galaxies.  We build geometric features from these sources and search
-for matching features in a large index.  Our approach will be
-described more fully in section \ref{ourapproach}.
-
-
-The blind astrometry problem is challenging for several reasons.
-First, a typical astronomical image covers a tiny fraction of the sky:
-the example image shown above covers about one millionth of the sky.
-Second, both the input image and the reference catalog have positional
-noise-- errors in the measured positions of sources due to turbulence
-of the atmosphere, distortion from the telescope optics, and image
-sensor noise.  Third, the input image measures an unknown portion of
-the electromagnetic spectrum (\emph{bandpass}).  In many cases filters
-have been used to isolate a narrow window of the spectrum.  This
-limits our ability to make use of the brightness of objects, since
-brightness in one band of the spectrum does not imply brightness (or
-even visibility) in another band.  Most reference catalogs measure
-brightness in only two or three bands.  Fourth, the input image and
-reference catalog have different effective exposure times, so a source
-visible in one may be below the detection threshold in the other.
-Finally, the input image can have nonlinear distortion due to the
-optical properties of the telescope.  These distortions are often
-modelled as polynomials up to fourth order, though higher orders are
-occasionally needed.
-
-
+solve, but one with practical implications for astronomers.
+
+
+For the purposes of astrometric calibration, we can think of the sky
+as a large two-dimensional surface: the stars are very distant, so our
+viewpoint is effectively fixed.  We are moving, as are the stars, but
+these motions are small relative to the precision at which we
+typically work.  The sky contains many stars, galaxies, and other
+astronomical sources.  The stars and distant galaxies are effectively
+point sources, while closer galaxies can be resolved.  Astrometric
+reference catalogs list the positions, motions, and brightnesses of
+these sources and serve as the ``ground truth'' or database of known
+(reference) objects.  The USNO-B1 catalog \cite{usnob, nomad}, for
+example, lists over one billion objects.  As many as a few percent of
+these are false detections or other artifacts \cite{barroncleaning},
+and some objects that should be visible are missing.
+
+
+The images to be recognized are subregions of the sky.  Image sizes
+range from nearly half the celestial sphere down to $10^{-7}$ of the
+area and smaller.  The input images measure unknown bands of the
+electromagnetic spectrum, and various nonlinear functions may have
+been applied to the pixel values.  We cannot rely on absolute
+brightness or color to recognize individual stars or galaxies.  At
+best we can hope that there is some positive correlation in the
+relative brightness ordering of objects in the image and the
+corresponding objects in our catalog.
+
+
+Blind astrometric calibration is an ideal task for exploring geometric
+ideas in pattern recognition.  Most celestial objects are effectively
+point sources, and can be found and localized to sub-pixel accuracy
+using relatively simple image-processing procedures.  But since the
+individual features are characterized only by their positions and
+brightnesses, we must examine collections of features in order to
+build distinctive patterns.  In \chapref{chap:techreport} we present
+\an, which applies the geometric hashing framework to the task of
+blind astrometric calibration.  An example of our results in shown in
+\figs \ref{fig:redgreen} and \ref{fig:onthesky}.
+
+
+\comment{
 An additional challenge in astrometry is that the input image and
 reference catalog may each have fictitious or missing sources.
-%
-\comment{ The most obvious effect is that the input image will in
-general have a different effective exposure time than the reference
-catalog, meaning that objects existing in one will be below the
-detection threshold in the other.  }
-%
 Extra sources can be due to planets, comets, satellites, or aircraft.
 Missing or poorly localized source can be due to imperfections in the
@@ -926 +945 @@
 distractors means that we can never assume that all the objects in the
 reference catalog will be contained in the image, or vice versa.
-
 
 There are several useful applications for a blind astrometry solver.
@@ -955 +973 @@
 control system.
 
-
-\subsubsection{Astrometry as a visual pattern recognition task}
-
 \begin{figure}
-\begin{center}
-\includegraphics[width=3.3in]{moon} \\
-\includegraphics[width=3.3in]{saturated}
-\end{center}
-\caption{Astronomical images with occlusion.  Top: the moon occludes the stars
-behind it (image copyright Johannes Schedler, \texttt{http://panther-observatory.com}).  Bottom:
-saturation and diffraction spikes due to bright objects can obscure nearby objects.}
+  \begin{center}
+        %\begin{tabular}{c@{\hspace{1pt}}c@{\hspace{1pt}}c}
+        \includegraphics[width=\figunit]{moon} \\
+        \includegraphics[width=\figunit]{saturated}
+        %\end{tabular}
+  \end{center}
+  \caption{Astronomical images with occlusion.  \captionpart{Top:} The
+moon, buildings, and mountains occlude the stars behind them (image
+copyright Johannes Schedler, \texttt{http://panther-observatory.com}).
+\captionpart{Bottom:} Saturation and diffraction spikes due to bright
+objects can obscure nearby objects.}
 \label{moon}
 \end{figure}
-
-The blind astrometry problem can be seen as a somewhat peculiar visual
-pattern recognition problem.  The images to be recognized are
-subregions of a large two-dimensional surface (for our purposes).  On
-this dark surface are many luminous objects, many of which are
-effectively point sources, and some of which are extended or nebulous.
-Unlike many object recognition tasks, we do not have ready access to
-these objects, so we must use existing information in the form of an
-astrometric reference catalog compiled by astronomers.  The number of
-objects listed in the catalog is of order $10^9$, but as many as a few
-percent are false detections or other artifacts, and some objects that
-should be visible are missing.
-
-%The cameras that generate the images which we are to recognize have unknown wavelength
-%bandpasses and various nonlinear functions may have been applied to the pixel values.
-%
-The input images measure unknown bands of the electromagnetic
-spectrum, and various nonlinear functions may have been applied to the
-pixel values.  We cannot therefore rely on absolute brightness or
-color.  At best we can hope that there is some positive correlation in
-the relative brightness ordering of objects in the image and the
-corresponding objects in our catalog.
-
 
 Since most celestial objects are effectively point sources, the
@@ -999 +995 @@
 are not distinctive.
 
-The scale of astronomical images can range from nearly half the total
-surface area of the celestial sphere down to $10^{-7}$ of the area or
-smaller.
-
 Although occlusion is typically not a problem in astronomical images,
 it does occasionally occur (see Figure \ref{moon}).  In addition,
@@ -1008 +1000 @@
 over a large region of the image, and this can cause nearby objects to
 be hidden.
-
-
-%In addition, the input image can have distortion, which results in changes
-
-
-The astrometry domain is clearly quite different from much of visual
-pattern recognition, where commonly-used features include edges,
-corners, curves, patches, and textured or textureless regions, where
-color, shape, and appearance are important, and where ``scale
-invariance'' rarely implies more than an order of magnitude.  The
-astrometry domain provides an excellent testbed for exploring
-geometric feature matching techniques, because little else is
-available.
-
-
-
-
+}
 
 

trunk/documents/theses/dstn/thesis.bib

@@ -7 +7 @@
     volume = {65},
     pages = {43--72}
+}
+
+@article{barroncleaning,
+  author={Jonathan T. Barron and Christopher Stumm and David W. Hogg and Dustin Lang and Sam Roweis},
+  title={CLEANING THE USNO-B CATALOG THROUGH AUTOMATIC DETECTION OF OPTICAL ARTIFACTS},
+  journal={The Astronomical Journal},
+  volume={135},
+  number={1},
+  pages={414-422},
+  url={http://stacks.iop.org/1538-3881/135/414},
+  year={2008},
 }
 

trunk/documents/theses/dstn/thesis.tex

@@ -1 @@
-%\documentclass{ut-thesis}
-\documentclass[draft,12pt]{ut-thesis}
+\documentclass{ut-thesis}
+%\documentclass[draft,12pt]{ut-thesis}
 
 \newcommand{\doctype}{chapter}