I am doing hand gesture recognition application, using OpenCV and single webcam.
Currently I'm basing on:
"Complex-Valued Levenberg-Marquardt Algorithm for Learning and Recognizing Various Hand Gestures"
and what I want to achieve is to connect nodes of hand skeleton (thinned hand contour), to get clean final skeleton where I can analyze angles and lengths (see picture).
I have no idea how can I separate the node points from others, and connect them like in the right picture.
Any help?
