0

So far I've found topics which describe how to compare to images by its preferences (saturation, blur,...) to see if they are identical.

However, what algorithm can help me to detect is an image (taken by a photo camera, mobile camera, etc.) contains a circle for example or a square?

In the database, I will have a list of objects (circle, square, polygons,...) and the algorithm should compare if any of these objects is present on a photo.

sandalone
  • 41,141
  • 63
  • 222
  • 338
  • 5
    http://stackoverflow.com/questions/5522771/how-to-detect-shapes-in-an-image – Maroun Feb 13 '14 at 07:56
  • 1
    You can use JavaCV for detecting shapes in images. IF you are interested in the algorithms you can go through OpenCV source code and wiki – Suvarna Pattayil Feb 13 '14 at 07:59
  • @VusP what about polygon detection? – sandalone Feb 13 '14 at 08:12
  • 1
    search hough transfrom – Vikram Bhat Feb 13 '14 at 08:13
  • switch to frequency domain via DFFT/DFT or DFCT/DCT and check the frequency distribution. if it is 'analog' like (smooth) then its most likely a photo (real world image) if it is pulse-like then it is drawed imaged (vector,cartoon,render,...) specific patterns can detect presence of basic shapes like rectangle,line,circle... this with combination of filters and image area selection and segmentation leads to more advanced computer vision algorithms ... – Spektre Feb 13 '14 at 08:47
  • 1
    Try [Viola Jones Object Detection](http://en.wikipedia.org/wiki/Viola%E2%80%93Jones_object_detection_framework) – amit Feb 13 '14 at 09:14

0 Answers0