cvFindFundamentalMat accuracy problem (Bug #1404)

Added by jremillard - over 13 years ago. Updated almost 13 years ago.

Status:Cancelled Start date:
Priority:Normal Due date:
Assignee:Ilya Lysenkov % Done:

0%
Category:calibration, 3d
Target version:2.4.0
Affected version: Operating System:
Difficulty: HW Platform:
Pull request:

Description

Upgraded from OpenCV 1.0, October 18, 2006 to 2.3.1. cvFindFundamentalMat in 2.3.1 has accuracy problem. Attached is a program that has the set of point that cause the error. The epi lines have an error in them of 0.05 pixels on this data set. Version 1.0 the epi lines were accurate to below 0.001 pixels. None of the points are bad, so the RANSAC part should not metter.

vxltestmain.cpp - test case (10.9 kB) jremillard -, 2011-09-30 11:45 pm

Associated revisions

Revision 0f4089f3
Added by Roman Donchenko over 11 years ago

Merge pull request #1404 from ilya-lavrenov:stdcall

History

#1
Updated by jremillard - over 13 years ago

Also, this is a 32 bit visual studio 2005, win 7 64 bit computer. I build opencv from source and did not use the windows setup kit.

#2
Updated by Alexander Shishkov about 13 years ago

  • Description changed from Upgraded from [[OpenCV]] 1.0, October 18, 2006 to 2.3.1. cvFindFundamentalMat... to Upgraded from OpenCV 1.0, October 18, 2006 to 2.3.1. cvFindFundamentalMat in ... More

#3
Updated by Alexander Shishkov almost 13 years ago

  • Priority changed from High to Normal
  • Target version deleted ()

#4
Updated by Alexander Shishkov almost 13 years ago

  • Assignee deleted (Vadim Pisarevsky)

#5
Updated by Ilya Lysenkov almost 13 years ago

  • Assignee set to Ilya Lysenkov

#6
Updated by Ilya Lysenkov almost 13 years ago

Thank you for the report and especially for the code.

The fundamental matrix is defined only up to scale. The computed matrices in your code differ only by a scale factor -1 within of a small error. So it is the same fundamental matrix. Please, make sure that you normalize a fundamental matrix in the same way before using it. If the problem is still occurs then re-open the issue and upload code to measure accuracy.

  • Status changed from Open to Cancelled

#7
Updated by Alexander Shishkov almost 13 years ago

  • Target version set to 2.4.0

Also available in: Atom PDF