0

Unfortunately, I was not able to find any solution to my question.

What I'm trying to do is to improve the results using the OpenCV-method cvFindChessboardCorners in order to be able to achieve a better camera calibration, because I think that this is the reason why I get poor results undistorting/rectifying images like in my question before (Question: Undistorting/rectify images with OpenCV).

So, what I want to know is, how I can improve the algorithm in order to be able to detect the chessboard corners in all images of this link: http://abload.de/img/cvfindchessboardcorneoxs73.jpg

Everything works fine, only in the two last images the chessboard could not be detected. Any help or advice or hint is very welcome!

Thanks in advance!

Community
  • 1
  • 1
VollNoob
  • 267
  • 4
  • 15
  • can you provide the original, unprocessed (problematic) single images in full resolution? – Micka Jan 31 '14 at 09:04
  • I got them from here: https://sites.google.com/site/scarabotix/ocamcalib-toolbox/ocamcalib-toolbox-download-page, you can download them at the bottom of that page, the filename is sampleimages.zip – VollNoob Feb 05 '14 at 08:03
  • ok... some single corners/squares aren't extracted by opencv's findChessboardCorners method (1 missing corner in the last image and 2 in the second last image). Computing a gradient magnitude image, the corners look 'ok', so maybe you can improve the result. I guess corners aren't extracted because there is too much white in some of the black squares from the light (so a thresholding problem). I'll try an approach: finding contours with gradient magnitude, then enhance the black squares by drawing filled contour. – Micka Feb 06 '14 at 11:03
  • Thanks very much! Now I'm trying to get to run the OpenCV sample-file "edge.cpp", but I somehow do not get it to work. I will keep you updatet on my progress – VollNoob Feb 07 '14 at 13:40
  • Unfortunately, calculating the edge map of the images did not help me to improve the results of cvFindChessboardCorners. – VollNoob Feb 25 '14 at 09:09
  • didn't find an easy solution either... sorry – Micka Feb 25 '14 at 10:40
  • No need to be sorry. I tried interpolating the given coordinates in order to calculate the intersection of the polynomials which I hoped would be the missing corner, but it did not work out too well... I also posed another question dealing with that. Thanks anyway :) – VollNoob Feb 25 '14 at 10:44

0 Answers0