Subpixel precision by LineSegmentDetector, Image Processing module (Feature #3890)


Added by Karel Knoest over 10 years ago. Updated over 9 years ago.


Status:Done Start date:2014-08-29
Priority:Normal Due date:
Assignee:- % Done:

100%

Category:imgproc, video
Target version:3.0
Difficulty:Easy Pull request:

Description

Dear Sir/Madam,

The LineSegmentDetector by Daniel Angelov in OpenCV 3.0 is doing an excellent job. The detector even seems to have some additional, unlocked potential.

Currently, the method LineSegmentDetector::detect returns the detected line segment coordinates with int precision (in a Vec4i object). The LSD algorithm and this implementation internally however work with subpixel precision. Only at the very last moment, the double results are cast to ints.

Could type of "lines" output parameter of LineSegmentDetector::detect be changed from from Vec4i to Vec4f? A quick local build shows good results.

Kind regards,
Karel Knoest


Associated revisions

Revision bd2be9c7
Added by Vadim Pisarevsky almost 10 years ago

Merge pull request #3890 from greensea:patch-1

History

Updated by Steven Puttemans over 10 years ago

I suggest you to make a pull request of your changes! This addition would be excellent.
More information on PR's can be found here: http://code.opencv.org/projects/opencv/wiki/How_to_contribute

  • Assignee changed from Vadim Pisarevsky to Karel Knoest
  • Status changed from New to Open

Updated by Karel Knoest about 10 years ago

  • Target version changed from 3.0-alpha to 3.0

Updated by Karel Knoest about 10 years ago

Pull request #3505.

  • % Done changed from 0 to 70

Updated by Abhijat Biswas almost 10 years ago

Shouldn't this issue be closed, seeing as the pull request was merged?

Updated by Steven Puttemans over 9 years ago

There you go!

  • Assignee deleted (Karel Knoest)
  • Status changed from Open to Done
  • % Done changed from 70 to 100

Also available in: Atom PDF