Mimas Camera Calibration
(Continuing ...) |
(Continuing ...) |
||
Line 17: | Line 17: | ||
\begin{pmatrix}m_{i1}\\m_{i2}\\m_{i3}\end{pmatrix} | \begin{pmatrix}m_{i1}\\m_{i2}\\m_{i3}\end{pmatrix} | ||
</math> | </math> | ||
− | with <math>\vec{m}^\prime_i=\widehat{\vec{m}}_i+\vec{\epsilon}_i</math>. | + | with <math>\vec{m}^\prime_i=\widehat{\vec{m}}_i+\vec{\epsilon}_i</math>, where |
+ | <math>\widehat{\vec{m}}_i</math> is the ideal position of the <math>i</math>th point in the camera-image. | ||
+ | <math>(\epsilon_{i1},\epsilon_{i2})^\top</math> is the zero-mean error in the observation of the <math>i</math>th point. | ||
− | Using <math>m_{i3}=m^\prime_{i3}=\widehat{m}_{i3}=1</math> and <math>\vec{h_i}^\top=\begin{pmatrix}h_{i1}&h_{i2}&h_{i3}\end{pmatrix},\ i\in\{1,2,3\}</math> the | + | Using <math>m_{i3}=m^\prime_{i3}=\widehat{m}_{i3}=1,\ \epsilon_{i3}=0</math> and <math>\vec{h_i}^\top=\begin{pmatrix}h_{i1}&h_{i2}&h_{i3}\end{pmatrix},\ i\in\{1,2,3\}</math> the |
− | model can be | + | model can be reformulated to |
<math> | <math> | ||
− | \begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\end{pmatrix}= | + | \begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\end{pmatrix}=\cfrac{1}{\vec{h_3}^\top\cdot\vec{m}_i}\, |
− | \cfrac{1}{\vec{h_3}^\top\cdot\vec{m}_i}\, | + | |
\begin{pmatrix}\vec{h_1}^\top\\\vec{h_2}^\top\end{pmatrix}\, | \begin{pmatrix}\vec{h_1}^\top\\\vec{h_2}^\top\end{pmatrix}\, | ||
\begin{pmatrix}m_{i1}\\m_{i2}\\1\end{pmatrix}+ | \begin{pmatrix}m_{i1}\\m_{i2}\\1\end{pmatrix}+ | ||
− | \vec{\ | + | \begin{pmatrix}\epsilon_{i1}\\\epsilon_{i2}\end{pmatrix} |
− | </math>. | + | </math> or <math> |
+ | \begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\end{pmatrix}\,\big(\vec{h_3}^\top\cdot\vec{m}_i\big)= | ||
+ | \begin{pmatrix}\vec{h_1}^\top\\\vec{h_2}^\top\end{pmatrix}\, | ||
+ | \begin{pmatrix}m_{i1}\\m_{i2}\\1\end{pmatrix}+ | ||
+ | \begin{pmatrix}\epsilon^\prime_{i1}\\\epsilon^\prime_{i2}\end{pmatrix} | ||
+ | </math> with | ||
+ | <math>\vec{\epsilon^\prime}_i=\big[\vec{h_3}^\top\cdot\vec{m}_i\big]\begin{pmatrix}\epsilon_{i1}\\\epsilon_{i2}\end{pmatrix}</math> | ||
+ | |||
+ | It is assumed, that <math>\vec{h_3}^\top\cdot\vec{m}_1,\,\vec{h_3}^\top\cdot\vec{m}_2,\,\ldots\approx\mathrm{const.}</math> and that the [http://en.wikipedia.org/wiki/Gauss-Markov_theorem Gauss-Markov theorem] can be applied with using <math>\vec{\epsilon^\prime}_i</math> as error-vectors. Therefore the [http://en.wikipedia.org/wiki/Least_squares miminum least-squares estimator] is the best linear estimator. | ||
[[Image:working.gif]] Under construction ... | [[Image:working.gif]] Under construction ... |
Revision as of 11:53, 26 June 2006
Calibration
Let <math>\vec{m}_i\in\mathbb{R}^3,\,i\in\{1,2,\ldots,N\}</math> be the homogeneous coordinate of the <math>i</math>th point on the planar calibration-object and let <math>\vec{m}^\prime_i\in\mathbb{R}^3</math> be the homogeneous coordinate of the corresponding pixel in the camera-image. Further let <math>(\mathbb{R}^n,\simeq)</math> be an equivalence relation defined by
<math> \vec{a}\simeq\vec{b}\ :\Leftrightarrow\ \exists\lambda\in\mathbb{R}/\{0\}:\,\lambda\,\vec{a}=\vec{b} </math>
If the camera-system does an ideal central projection (e.g. no distortion), the projective transformation (the Homography) can be modelled using <math>\mathcal{H}\in\mathbb{R}^{3\times 3}</math> as follows
<math> \widehat{\vec{m}}_i\simeq\mathcal{H}\vec{m}_i </math> or more elaborately <math> \lambda\,\begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\\\widehat{m}_{i3}\end{pmatrix}= \begin{pmatrix}h_{11}&h_{12}&h_{13}\\h_{21}&h_{22}&h_{23}\\h_{31}&h_{32}&h_{33}\end{pmatrix}\, \begin{pmatrix}m_{i1}\\m_{i2}\\m_{i3}\end{pmatrix} </math> with <math>\vec{m}^\prime_i=\widehat{\vec{m}}_i+\vec{\epsilon}_i</math>, where <math>\widehat{\vec{m}}_i</math> is the ideal position of the <math>i</math>th point in the camera-image. <math>(\epsilon_{i1},\epsilon_{i2})^\top</math> is the zero-mean error in the observation of the <math>i</math>th point.
Using <math>m_{i3}=m^\prime_{i3}=\widehat{m}_{i3}=1,\ \epsilon_{i3}=0</math> and <math>\vec{h_i}^\top=\begin{pmatrix}h_{i1}&h_{i2}&h_{i3}\end{pmatrix},\ i\in\{1,2,3\}</math> the model can be reformulated to <math> \begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\end{pmatrix}=\cfrac{1}{\vec{h_3}^\top\cdot\vec{m}_i}\, \begin{pmatrix}\vec{h_1}^\top\\\vec{h_2}^\top\end{pmatrix}\, \begin{pmatrix}m_{i1}\\m_{i2}\\1\end{pmatrix}+ \begin{pmatrix}\epsilon_{i1}\\\epsilon_{i2}\end{pmatrix} </math> or <math> \begin{pmatrix}\widehat{m}_{i1}\\\widehat{m}_{i2}\end{pmatrix}\,\big(\vec{h_3}^\top\cdot\vec{m}_i\big)= \begin{pmatrix}\vec{h_1}^\top\\\vec{h_2}^\top\end{pmatrix}\, \begin{pmatrix}m_{i1}\\m_{i2}\\1\end{pmatrix}+ \begin{pmatrix}\epsilon^\prime_{i1}\\\epsilon^\prime_{i2}\end{pmatrix} </math> with <math>\vec{\epsilon^\prime}_i=\big[\vec{h_3}^\top\cdot\vec{m}_i\big]\begin{pmatrix}\epsilon_{i1}\\\epsilon_{i2}\end{pmatrix}</math>
It is assumed, that <math>\vec{h_3}^\top\cdot\vec{m}_1,\,\vec{h_3}^\top\cdot\vec{m}_2,\,\ldots\approx\mathrm{const.}</math> and that the Gauss-Markov theorem can be applied with using <math>\vec{\epsilon^\prime}_i</math> as error-vectors. Therefore the miminum least-squares estimator is the best linear estimator.
Other symbols: <math> \vec{0}, \neq, ||\mathcal{A}||, \mathcal{A}\in\mathbb{C}^{3\times 3}, i\in\mathbb{N}_0,0\approx 1, \begin{pmatrix}1-\lambda&0\\0&1-\lambda\end{pmatrix}, \mathbb{R}^3, i\in\{1,2,\ldots,n\}</math>
<math> \begin{pmatrix}1-\lambda&0&\cdots&\cdots&0\\0&1-\lambda&\ddots&&\vdots\end{pmatrix},V^*,a*b, \sigma, \Sigma, \lambda, \Lambda</math>
<math> \sum_{i=1}^n p^i, \prod_{i=1}^{+\infty}a_i, \cos(\omega), \widehat{i}=\mathrm{argmax}_{i\in\mathbb{N}}f(i), a\simeq b </math>