Augmented Reality: Just marker recognition

Question

Think about the usual augmented reality implementations: there's one or more marker, the software recognize the markers and usually shows something on top of them.

I'm interested just in the first part. I need an algorithm, component, framework or class to determine if a given marker is present or not in a photo.

Where can I start from? I prefer a component for Java or C++, but I'm not limited by these languages.

Answer 1

OpenCV for C++ (also has Python bindings) would be a good place to start.

Answer 2

You essentially need to track an object through a video. There are several ways to do that with OpenCV which are explained in their page on Motion Analysis and Object Tracking.

Answer 3

I'm not sure this if this is helpful or not, but for Java, you might look at ZXing. It is a barcode reader library, but the task of reading QR codes amounts to recognizing 3-4 markers in the barcode. You might find its approach reusable.

Answer 4

I guess you should see opencv' detectors: Feature Detection (there is example, find_obj.cpp sample shipped with opencv package). The most popular algorithm is described and implemented in ARToolKit (just google it).