Ask Your Question
2

What kind of features that can be extracted from binary image

asked 2013-04-04 14:47:53 -0500

Heshan Sandeepa gravatar image

I am doing OCR application in c++ and opencv 2.4.4. I use LIBSVM for classification. In order to classification I have extracted some features like HuMoments, Area of Black-pixels, Ratio between black and white pixels, Ratio of image binding box width and height, Number intersections through the image and Number of end points of the image. In addition to those, i have read that we can use histogram based features and image profile based features. But i have no idea what are they. Can anyone give some examples about them or can anyone say that, what are the other kind of features that i can use to make the feature vector more rich.

edit retag flag offensive close merge delete

3 answers

Sort by ยป oldest newest most voted
5

answered 2013-04-05 01:13:58 -0500

mrgloom gravatar image
edit flag offensive delete link more
4

answered 2013-04-04 17:37:26 -0500

Guanta gravatar image

It's more an addition to Stevens answer:

  • Histogram based features are e.g. HOG features, or Histograms of LBPs.
  • Profile features are computed from the outer boundary of your letter / word in a specific direction, e.g. left profile: go from the left bounding box border to the right until you hit your first pixel and do this for each row (thus a proper binarization is important) and you'll get a left profile. You can do the same for the other directions too (top to bottom, bottom to top, left to right, right to left). From these profiles you can then compute several statistics (mean, std-dev., etc.).
edit flag offensive delete link more
2

answered 2013-04-04 15:43:11 -0500

A link to color histogram matching :

http://facweb.cs.depaul.edu/research/techreports/tr06-010.pdf

About image profile, guess they are talking about interesting keypoints to detect in the image or about the gradient image. Information can be found on the following links:

Altough remember that feature point extractors do not work well with binary images. They need more information to find enough points.

edit flag offensive delete link more
Login/Signup to Answer

Question Tools

1 follower

Stats

Asked: 2013-04-04 14:47:53 -0500

Seen: 2,369 times

Last updated: Apr 05 '13