Given is a Equation in the Form

$A_{i}\underline{r}=0$ like:

$\begin{bmatrix}\underline{0}^T & -\underline{X}_{i}^T & y_{i}\underline{X}_{i}^T\\\underline{X}_{i}^T & \underline{0}^T &-x_{i}\underline{X}_{i}^T \end{bmatrix} \begin{bmatrix} \underline{r}_{1} \\ \underline{r}_{2}\\ \underline{r}_{3}\end{bmatrix} = \begin{bmatrix}\underline{0}\end{bmatrix}$

$A$ can be extended to be a 12 x 12 matrix and $\underline{r} $ should be a 4x1 matrix

I assume that $\underline{r}^T$ is like [a, b, c, 1]. How can I solve this equation with numpy? It is recommended to solve the Equation using SVD, but I doubt that this is the right way.

  • $\begingroup$ $A$ is $3 \times 12$ and $r$ is $4 \times 1$ you say. Are you sure ? $\endgroup$ – Ahmad Bazzi Sep 5 '18 at 13:11
  • $\begingroup$ You're right. I've corrected it. But r should be 4x1 $\endgroup$ – Bryan McGill Sep 5 '18 at 13:27
  • $\begingroup$ $A$ is $12 \times 12$ and $r$ is $4 \times 1$. Are you sure ? $\endgroup$ – Ahmad Bazzi Sep 5 '18 at 13:47
  • $\begingroup$ This is the formula I read in the paper. A is actually 2x 12 but it is mentioned, that A should be extended to A 12 by 12: See p.52 [link](appliedmaths.sun.ac.za/TW792/chapter5.pdf) $\endgroup$ – Bryan McGill Sep 5 '18 at 13:54
  • $\begingroup$ The $r$s here are the transposed rows of a $4\times3$ camera matrix, not homogeneous coordinates of points. They do not all end in $1$. In theory, the solution set is just the null space of the coefficient matrix. In practice, your system will be overdetermined and, as the text book says, computing the SVD is a fairly efficient way of estimating a solution. $\endgroup$ – amd Sep 5 '18 at 22:31

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.