Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[Ch7, 7.7.3] Questions #39

Open
nicolasrosa opened this issue Mar 22, 2021 · 2 comments
Open

[Ch7, 7.7.3] Questions #39

nicolasrosa opened this issue Mar 22, 2021 · 2 comments

Comments

@nicolasrosa
Copy link
Collaborator

nicolasrosa commented Mar 22, 2021

Hello @gaoxiang12,

Sometimes I get very confused about the notations used. For example, in topic 7.7.3, I believe that for being consistent with those presented in chapter 5, we can do the following notation equivalences:

CodeCogsEqn (1)

Am I correct?

  1. When you say "3D space points P", are you referring to Pw, correct?

  2. About R, t, in this topic, is Rcw and tcw, correct?

  3. In equation 7.35, T should be T*, right?

  4. In equation 7.43, why there is a negative sign? Have you considered the following?
    CodeCogsEqn (2)

    If yes, should the equation 7.36 be reversed? Numerically, it shouldn't change because the resulting value is squared in the least-squares minimization, but I would change it for didactic reasons since until then you haven't presented the residual definition.

UPDATE: It doesn't need to answer question 4. At the end of the reading, you said the definition of the error.

@nicolasrosa nicolasrosa changed the title [Ch7, 7.7.3] Notation changes [Ch7, 7.7.3] Questions Mar 22, 2021
@gaoxiang12
Copy link
Owner

Hi @nicolasrosa ,

  1. Yes, and since we don't explicitly distinguish the homogeneous/non-homogeneous coordinates, we use a simple notation Pw in the text. Sometimes we multiply it with R and it is a 3D vector, and sometimes we multiply it with T so it is a 4D vector at that time. I think it makes the text more fluent. Otherwise, we may add a lot of superscripts to the equations.
  2. Yes. We use world-to-camera transform as default. But if we consider a two-view problem, we may use T12 or T21 as the estimated variable.
  3. Not sure. What do you mean by T*? I basically use the star superscript for the optimal solution like (7.36).

@nicolasrosa
Copy link
Collaborator Author

nicolasrosa commented Mar 23, 2021

  1. Nice, this confirms what I was suspecting. The only problem is that, so far in the book, P = [X, Y, Z]^T has been used to represent the 3D space points described in the {camera} frame, like in the topic 7.3.1 Epipolar Geometry on page 157.

    Screenshot from 2021-03-23 15-26-24

    Considering your next answer, it becomes clear that for the PnP problem, P now represents the 3D space points in {world} frame.

  2. This answer clarified everything for me!

  3. Yes, I just said that because without the * we cannot differentiate the following equations, right?

    CodeCogsEqn

Thank you, your answers really helped me!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants