0

I have a straightforward (I guess) problem to solve with opencv:

  1. I have 3 non-colinear Points in the 3D system given in cm:
Point A = Point(.., .., ..);
Point B = Point(.., .., ..);
Point C = Point(.., .., ..);
  1. I have the respective 2D Points relative to an image with height H and width W, everything given in pixels:
Point A_proj = Point(.., ..);
Point B_proj = Point(.., ..);
Point C_proj = Point(.., ..);
  1. The camera from which the image was grabbed has focal length f 3.67 mm

How can I estimate the 6-dof camera position (translation and rotation) from those data using OpenCV? Is that possible?

Example

This is the coordinates of points in 3D system:

real world point coordinates

This is the a 640x480 image taken from some camera pose I want to estimate:

projection perspective

Duloren
  • 2,395
  • 1
  • 25
  • 36
  • Can you clarify whether you know intrinsic camera calibration -- focal length in pixels (alternatively, camera field-of-view), optical center, and distortion parameters? This is a [Perspective-n-Point](https://en.wikipedia.org/wiki/Perspective-n-Point) problem, but the solution assumes that camera calibration is known. – vasiliykarasev Jul 26 '20 at 22:15
  • @vasiliykarasev I think I can convert the focal length from mm to pixels, so, feel free to consider this value as given. For simplicity, we can take the camera as calibrated already hence we have the intrinsecs matrix at hand. – Duloren Jul 26 '20 at 22:39

0 Answers0