📄 help_imare.html
字号:
<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="GENERATOR" content="Microsoft FrontPage 5.0">
<title>IMARE Toolbox Help</title>
</head>
<body text="#000099" bgcolor="#FFFFFF" link="#006600" vlink="#009900" alink="#FF0000">
<center><b><font color="#3333FF"><font size=+1>IMARE registration toolbox</font></font></b><p><font color="#000099"> <a href="mailto:zitova@utia.cas.cz">Barbara Zitova</a>.</font>
</p>
<p> </p>
</center>
<p align="justify"><font color="#000080">This toolbox was created for the study
of moment invariants and registration algorithms, which use them. It can be
easily extended using other types of invariants, feature detection algorithms
and registrations methods. This set of tools was not meant to be the solution
for real applications. It allows to better understand the behavior of various
moment invariants on various image data. </font></p>
<p><br>
<center><table WIDTH="100%" >
<tr>
<td>
<ul>
<li>
<b><font color="#72BF64"><font size=+1><a href="#notes">General notes on
used image registration algorithms</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#IO">Data input/output</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#viz">Visualization</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#actions">Actions and parameter
setting</a></font></font></b></li>
<ul>
<li>
<b><font color="#72BF64"><font size=+1><a href="#CPdet">Control point detection</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#invc">Invariant computation</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#match">Matching</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#reg">Registration</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#eval">Evaluation</a></font></font></b></li>
<li>
<b><font color="#72BF64"><font size=+1><a href="#ncc">Normalized cross-correlation</a></font></font></b></li>
<li>
<font size="+1"><b><a href="#auto">Phase correlation</a></b></font></li>
</ul>
<li>
<b><font size=+1 color="#72BF64"><a href="#info">Info</a></font></b></li>
</ul>
</td>
<td>
<center><img SRC="imare.jpg" NOSAVE height=307 width=375></center>
</td>
</tr>
</table></center>
<p>
<hr SIZE=3 WIDTH="100%">
<p><a NAME="notes"></a><font color="#009900"><font size=+1>General notes
on image registration algorithms</font></font>
<p><font color="#000099">Two basic approaches used for image registration
are implemented in IMARE registration toolbox. <b>Area-based </b><a href="#ncc">normalized
cross-correlation</a> (NCC) and <a href="#auto">phase correlation</a> and <b>feature-based</b> <a href="#reg">registration
by means of moment invariants</a>.</font>
<br>
<p><font color="#000099"> The <b>area-based</b> methods
can be used for shifted images, phase correlation for data from different
modalities too. <b>Feature-based</b></font> method can be applied on mutually
shifted and rotated images (Hu invariants, rotational invariants), affinely
distorted data (affine invariants), and rotated, shifted and blurred images
(combined blur invariants).<p><b>A</b><font color="#000099"><b>rea-based</b> methods: the window
is defined in the sensed image by </font><b><font color="#993300"><Window
selection> </font></b><font color="#000099">and computes NCC/PHC for all possible
shifts of the window across the Reference image by </font><b><font color="#993300"><Normalized
CC> / <Phase correl></font></b><font color="#000099">.</font>
<p><font color="#000099"> <b>Feature-based</b></font> method<font color="#000099"> finds the
registration parameters by means of detected control points and their mutual
correspondence. The correspondence is found in the feature space of moment invariants.The sequence of method steps is as follows</font>
<ul>
<ul>
<ul>
<li>
<font color="#000099"><a href="#CPdet">control point detection (CPs)</a>
</font><b><font color="#993300"><CP detection></font></b></li>
<li>
<font color="#000099"><a href="#INVc">computation of invariants for CPs </a>
</font><b><font color="#993300"><INV comp></font></b></li>
<li>
<font color="#000099"><a href="#invc">choice of invariants</a>
</font><b><font color="#993300"><Choice of invariants></font></b></li>
</ul>
</ul>
<ul>
<ul>
<li>
<font color="#000099"><a href="#match">CP matching </a>
</font><b><font color="#993300"><Two best></font></b></li>
<li>
<font color="#000099"><a href="#reg">final registration </a>
</font><b><font color="#993300"><Registration></font></b></li>
</ul>
</ul>
</ul>
<p><br><font color="#000099"> In both cases, resulting
registration can be evaluated by means of several methods - difference
image, averaged image etc. </font><b><font color="#993300"><Evaluation></font></b>
<p>
<hr SIZE=3 WIDTH="100%">
<br><a NAME="IO"></a>
<br><b><font color="#72BF64"><font size=+1>Data input/output</font></font></b>
<p><font color="#000099">IMARE toolbox supports input and output of images
(Reference, Sensed, transformed Sensed, CPs from Reference and Sensed images)
and computed values (positions of CPs, used parameters for image registration,
computed invariants, found registration parameters). Reference and Sensed
images have to be imported every time, invariants and control points can
be either imported or computed.</font>
<br>
<p><b><font color="#000099">Input of images to be registered</font></b>
<br><b><font color="#993300">
<Reference image></font></b>
<br><b><font color="#993300">
<Sensed image></font></b>
<br><font color="#000099">They can be in PGM, BMP, JPEG, TIFF or PCX format.</font>
<br>
<p><b><font color="#000099">Input of detected CP points (from previous
session)</font></b>
<br><b><font color="#993300">
<CP sets></font></b>
<br><font color="#000099">They can be in PGM (0-1 matrix with 1s at the
position of CPs) or MAT format (x and y coordinates).</font>
<br>
<p><b><font color="#000099">Input of computed invariants (from previous
session)</font></b>
<br><b><font color="#993300">
<Invariants></font></b>
<br><font color="#000099">They should be in MAT format. The file
should consists of</font>
<ul>
<ul>
<ul>
<li>
<font color="#000099">INV
values of all computed invariants</font></li>
<li>
<font color="#000099">INVS
values of choosen invariants for registration procedure</font></li>
<li>
<font color="#000099">filenm
name of the image file, used for invariant computation</font></li>
<li>
<font color="#000099">filename
name of the file with CPs used for invariant computation</font></li>
<li>
<font color="#000099">tp
type of computed invariants</font></li>
<li>
<font color="#000099">rd
radius of the circular neighborhood used for invariant computation</font></li>
<li>
<font color="#000099">rdd <a NAME="cperr"></a>
threshold for the expected error in CP detection in the Sensed image. Here,
invariants in the Sensed image are computed</font></li>
</ul>
<font color="#000099">
not just for the detected CP but as well for "close" points. The number
of "close" poin is set by rdd. [rdd = 1: the pink</font>
<br><font color="#000099">
pixel points are included, rdd = 2: pink and green, rdd = 3: pink, green
and blue].</font>
<br>
<center><table BORDER=2 >
<tr BGCOLOR="#3366FF">
<td>
<center><font color="#3366FF">**</font></center>
</td>
<td><font color="#3366FF">**</font></td>
<td><font color="#3366FF">**</font></td>
<td><b><font color="#FFFFFF">rdd=3</font></b></td>
<td><font color="#3366FF">**</font></td>
<td><font color="#3366FF">**</font></td>
<td><font color="#3366FF">**</font></td>
</tr>
<tr BGCOLOR="#99FF99">
<td BGCOLOR="#3366FF"><font color="#3366FF">**</font></td>
<td><font color="#99FF99">**</font></td>
<td><font color="#99FF99">**</font></td>
<td><b><font color="#3366FF">rdd=2</font></b></td>
<td><font color="#99FF99">**</font></td>
<td><font color="#99FF99">**</font></td>
<td BGCOLOR="#3366FF"> <font color="#3366FF">**</font></td>
</tr>
<tr BGCOLOR="#FFCC99">
<td BGCOLOR="#3366FF"> <font color="#3366FF">**</font></td>
<td BGCOLOR="#99FF99"><font color="#99FF99">**</font></td>
<td> <font color="#FFCC99">**</font></td>
<td><b><font color="#3366FF">rdd=1</font></b></td>
<td> <font color="#FFCC99">**</font></td>
<td BGCOLOR="#99FF99"> <font color="#99FF99">**</font></td>
<td BGCOLOR="#3366FF"> <font color="#3366FF">**</font></td>
</tr>
<tr>
<td BGCOLOR="#3366FF"> <font color="#3366FF">**</font></td>
<td BGCOLOR="#99FF99"> <font color="#99FF99">**</font></td>
<td BGCOLOR="#FFCC99"> <font color="#FFCC99">**</font></td>
<td ALIGN=CENTER BGCOLOR="#990000"><b><font color="#FFFFFF"><font size=+1>CP</font></font></b></td>
<td BGCOLOR="#FFCC99"> <font color="#FFCC99">**</font></td>
<td BGCOLOR="#99FF99"> <font color="#99FF99">**</font></td>
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -