In this paper, we present a method for hand gesture recognition using Microsoft Kinect sensor. Kinect allows capturing dense, and three dimensional scans of an object in real time. We propose a combination of modelling and learning approach for hand gesture recognition. We use Kinect depth feature for background segmentation of hand gesture images captured with Kinect. Image processing techniques are employed to find contour of segmented hand images. Then we calculate convex hull and convexity defects for this contour. We are using contour area and convexity defects as features for classification. We classify the gestures using naïve Bayes classifier. We have considered five hand gestures classes i.e. To show using one, two, three, four, and five fingers one by one. We implemented and tested this algorithm for 15 images of each class. It gives a correct classification rate of 100\%. © 2014 IEEE.