Analysis and implementation of simple gesture recognition algorithms using MATLAB

Gesture recognition is an easy and hassle free mode of human computer interaction. This paper analyzes and implements four techniques which are distance invariant pixel count algorithm, peak counting algorithm, objects counting algorithm and template matching. The implementation of these techniques in this paper is limited to finding the finger count in moderate to complex backgrounds. Some modifications are proposed which make the algorithms invariant to rotation, translation and scale of hand. These techniques are shown to work fairly well in cluttered and noisy backgrounds. They are demonstrated on real imagery and the results are compared on the basis robustness, accuracy and background in-variance.