Hey everbody,
I'm trying to do an icon-recognition program and already have achieved quite some thing with phashes for images. Meaning, if I have an image of an icon, I can quite fast match it to its original.
Now I'd like to do this for iOS-homescreens. Meaning: I have an iOS homescreen like so http://static.squarespace.com/static/5008676d84aeae82b8acdd8c/t/52578ca9e4b033e8d6d08be6/%3Fformat%3D500w and would like to know every position and size where I can find an icon.
In my imagination this seems a fairly easy problem, considering what opencv can do. But I couldn't figure out a good way working on it for one day.
I tried training a haar cascade but didn't get anything good. I also tried finding contours, but got to many results.
What would be a good way to approach this problem?
Thx for any help!
