-1

I am working on a image processing project which is related to currency recognition. What i am going to do is just load all currency images to an array and compare them one by one with the unknown currency that the user input to the system and finally to give the unknown currency value.To compare images I hope to use opencv SURF function. what is the best way to do this using opencv and java ? can anyone please let me know the best workflow for this?

Chathuranga Dulaj
  • 33
  • 1
  • 1
  • 4

1 Answers1

0

I made a similar project too. Since you require an algorithm that would be rotation and size invariant you can't use template matching. It would be unnecessary wastage of time and efforts if you train cascades.

I tried using OCR but that made things too complicated.

I think the best method to be used is either SIFT or SURF. So I think you should experiment a bit with SIFT too and then compare the accuracy to SURF.

If you want the detection to be quicker you can use FLANN based matcher.http://docs.opencv.org/3.1.0/dc/dc3/tutorial_py_matcher.html#gsc.tab=0

Arjun
  • 3
  • 1
  • 6