如何从特征点识别预定义手势

Question

I'm working on a hand gesture recognition project. Till now I'm able to detect the centre of the hand and able to track the hand for some consecutive frames and I'm able to get a list of points Like

1

2

Now I want to recognize the path as a gesture like for #1 RIGHT for #2 CIRCLE How should I do it ? and It should include a method to add other gestures to . Can I use SVM for this purpose? I feel it can be done by using FSM but I can't get how to implement it . I'm using Python and Opencv . Thanx in Advance!

Answer 1

There are various ways to approach this, but I believe that the easiest is to use a template matching approach.

For each gesture, have a sample that you compare to, and the result is simply the one most resembling the current sample.

For the comparison between a sample and a template, a good and simple to implement algorithm is the Dynamic Time Warping (DTW) https://en.wikipedia.org/wiki/Dynamic_time_warping