<?xml version="1.0"?>
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
	<id>http://debianws.lexgopc.com/wiki143/api.php?action=feedcontributions&amp;feedformat=atom&amp;user=92.76.185.7</id>
	<title>wiki143 - User contributions [en]</title>
	<link rel="self" type="application/atom+xml" href="http://debianws.lexgopc.com/wiki143/api.php?action=feedcontributions&amp;feedformat=atom&amp;user=92.76.185.7"/>
	<link rel="alternate" type="text/html" href="http://debianws.lexgopc.com/wiki143/index.php?title=Special:Contributions/92.76.185.7"/>
	<updated>2026-05-14T16:21:12Z</updated>
	<subtitle>User contributions</subtitle>
	<generator>MediaWiki 1.43.1</generator>
	<entry>
		<id>http://debianws.lexgopc.com/wiki143/index.php?title=Reprojection_error&amp;diff=5718129</id>
		<title>Reprojection error</title>
		<link rel="alternate" type="text/html" href="http://debianws.lexgopc.com/wiki143/index.php?title=Reprojection_error&amp;diff=5718129"/>
		<updated>2023-12-18T14:06:09Z</updated>

		<summary type="html">&lt;p&gt;92.76.185.7: Reprojection error has nothing to do with old phones, old cameras or image artifacts, this was either vandalism or a case of mistaken identity.&lt;/p&gt;
&lt;hr /&gt;
&lt;div&gt;{{Inline|date=July 2014}}&lt;br /&gt;
The &#039;&#039;&#039;reprojection error&#039;&#039;&#039; is a geometric error corresponding to the image distance between a projected point and a measured one. It is used to quantify how closely an estimate of a 3D point &amp;lt;math&amp;gt;\hat{\mathbf{X}}&amp;lt;/math&amp;gt; recreates the point&#039;s true projection &amp;lt;math&amp;gt;\mathbf{x}&amp;lt;/math&amp;gt;. More precisely, let &amp;lt;math&amp;gt;\mathbf{P}&amp;lt;/math&amp;gt; be the [[camera matrix|projection matrix]] of a [[pinhole camera|camera]] and &amp;lt;math&amp;gt;\hat{\mathbf{x}}&amp;lt;/math&amp;gt; be the image projection of &amp;lt;math&amp;gt;\hat{\mathbf{X}}&amp;lt;/math&amp;gt;, i.e. &amp;lt;math&amp;gt;\hat{\mathbf{x}}=\mathbf{P} \, \hat{\mathbf{X}}&amp;lt;/math&amp;gt;. The reprojection error of &amp;lt;math&amp;gt;\hat{\mathbf{X}}&amp;lt;/math&amp;gt; is given by &amp;lt;math&amp;gt;d(\mathbf{x}, \, \hat{\mathbf{x}})&amp;lt;/math&amp;gt;, where &amp;lt;math&amp;gt;d(\mathbf{x}, \, \hat{\mathbf{x}})&amp;lt;/math&amp;gt; denotes the [[Euclidean distance]] between the image points represented by vectors &amp;lt;math&amp;gt;\mathbf{x}&amp;lt;/math&amp;gt; and &amp;lt;math&amp;gt;\hat{\mathbf{x}}&amp;lt;/math&amp;gt;.&lt;br /&gt;
&lt;br /&gt;
Minimizing the reprojection error can be used for estimating the error from point correspondences between two images. Suppose we are given 2D to 2D point imperfect correspondences &amp;lt;math&amp;gt;\{\mathbf{x_i} \leftrightarrow \mathbf{x_i}&#039;\}&amp;lt;/math&amp;gt;. We wish to find a [[homography (computer vision)|homography]] &amp;lt;math&amp;gt;\hat{\mathbf{H}}&amp;lt;/math&amp;gt; and pairs of perfectly matched points &amp;lt;math&amp;gt;\hat{\mathbf{x_i}}&amp;lt;/math&amp;gt; and &amp;lt;math&amp;gt;\hat{\mathbf{x}}_i&#039;&amp;lt;/math&amp;gt;, i.e. points that satisfy &amp;lt;math&amp;gt;\hat{\mathbf{x_i}}&#039; = \hat{H}\mathbf{\hat{x}_i}&amp;lt;/math&amp;gt; that minimize the reprojection error function given by&lt;br /&gt;
:&amp;lt;math&amp;gt; \sum_i d(\mathbf{x_i}, \hat{\mathbf{x_i}})^2 + d(\mathbf{x_i}&#039;, \hat{\mathbf{x_i}}&#039;)^2&amp;lt;/math&amp;gt;&lt;br /&gt;
So the correspondences can be interpreted as imperfect images of a world point and the reprojection error quantifies their deviation from the true image projections &amp;lt;math&amp;gt;\hat{\mathbf{x_i}}, \hat{\mathbf{x_i}}&#039;&amp;lt;/math&amp;gt;&lt;br /&gt;
&lt;br /&gt;
==References==&lt;br /&gt;
&lt;br /&gt;
*{{cite book |&lt;br /&gt;
author=Richard Hartley and Andrew Zisserman |&lt;br /&gt;
title=Multiple View Geometry in computer vision |&lt;br /&gt;
publisher=Cambridge University Press|&lt;br /&gt;
year=2003 |&lt;br /&gt;
isbn=0-521-54051-8}}&lt;br /&gt;
&lt;br /&gt;
[[Category:Geometry in computer vision]]&lt;/div&gt;</summary>
		<author><name>92.76.185.7</name></author>
	</entry>
</feed>