Shyam, it is not necessary to

Shyam, it is not necessary to compare depth images.

If you, for instance want to have a gesture which is showing al 5 fingers with the palm facing the hand then you would look for a hand which has all it's fingers open and the palm normal vector facing the camera. or you could have a gesture which is activated when the hand is held in one place for (example) 2 seconds, which means that (for example again) the palm coordinates do not change alot over that timespan (just some simple calculations)

write a program which lets you see all the parameters available and then show your hand the way you want your gesture to be. take the most important parameters, add (something like) +/- 20% and there you go.

this is just how i am doing my gestures and probably not very good, but i hope it gives you a basic understanding.