2

I have a pair of images and the corresponding points taken from an uncalibrated camera. I also have the calibration matrix of the camera. Is it possible to multiply the 2D image coordinates by the inverse of the calibration matrix and do the reconstruction process like the calibrated case ? (Essential matrix and etc)

Moeen Eragh
  • 53
  • 1
  • 3
  • have you read [this](http://opencv.willowgarage.com/documentation/python/camera_calibration_and_3d_reconstruction.html) and [this](http://www.cs.unc.edu/~marc/tutorial/node37.html)? – bitWorking Jun 29 '13 at 19:47

1 Answers1

0

Yes, you can do this like you described. To be more exact, you can multiply your measured image points in homogenous coordinates with the inverse of your calibration matrix from the left hand side.

The coordinates of this operation are than in normalized camera coordinates and you can compute the essential matrix. Another possibility is to compute the fundamental matrix with the measured coordinates, for instance, with the normalized 8-point algorithm and compute the essential matrix afterwards like given in this talk (slide 6 about the epipolar constraint).

Make sure that the two of singular values of the essential matrix are equal and the third is zero.

who9vy
  • 1,091
  • 9
  • 19