Geek Answers Handbook

Logo detection with OpenCV

I am trying to implement an Easter egg in a mobile application that I am working on. These Easter eggs will be triggered when a camera logo is detected. The logo I'm trying to detect is as follows: Halifax logo .

I'm not quite sure that the best way to get close to this is that I'm pretty new to computer vision. I am currently finding horizontal edges using the Canny algorithm. Then I find the line segments using the Hough probabilistic transform. The conclusion of this is as follows (blue lines represent line segments detected by the Hough probabilistic transformation):

halifax logo post detection

The next step I was about to take would be to search for a group of approximately 24 lines (set in a rectangular rectangle), each line should be approximately the same length. I would use these two signals to indicate the potential presence of the logo. I understand that this is probably a very naive approach and would welcome suggestions on how to better define this logo in a more reliable way?

thanks

+8
opencv computer-vision image-recognition
source share
3 answers

You can go with SIFT using Rob Hess SIFT Library . It uses OpenCV as well as pretty fast. I think this is easier than your current way of approaching logo detection :)

Try also look for SURF , which claims to be faster and wider than SIFT. This Object Detection Guide will help you.

+4
source share

You can simply use LogoGrab technology. It is best there and offers all kinds of APIs (both mobile and HTTP). http://www.logograb.com/technologyteam/

+2
source share

I'm not quite sure that you will find such features in the logo to use the SIFT / SURF approach. As an alternative, you can try to train a clumsy classifier of objects and use it to detect a logo, just like opencv for face recognition.

0
source share

More articles:


All Articles