I am new to openframeworks, kinect, openCV, projections - you name it!
The aim of this quest is to match coordinates from an object (f.e a box) tracked with Kinect, to a digital projection that will track the objects in view. much like painting digitally the side of a box which is also moving within the field of view of the Kinect camera.
I’ve been reading theory on projections and it seems that homography is more or less what I need to be reading upon. Is that correct?
in practice I am attempting to figure out which addon is most suitable.
It looks like opencv2 exposes the functions findHomography() and perspectiveTransform() from http://docs.opencv.org/doc/tutorials/features2d/feature-homography/feature-homography.html are of interest. I am not sure if this 3x3 or 4x4 homography matrix can account for titling of the Kinect.
Additionally, I am completely fuzzy about the use of the term “Perspective distortion removing” used.
Do I need to consider un-distorting my camera view?
Also I wonder what is the path of least pain to achieve what I need: ofxOpenCV2? ofxCv? or the ofxCv2 fork? and what about ofxCamaraLucida? I am oblivious as to the relevant features that each addon is trying to implement so as to decide which one suits me best…
May I please ask for a kick start or some links / info regarding what is available?
Thank you in advance for your help