ABSTRACT

Gesture recognition is very much important for many applications of vision based systems but it is very difficult to recognize due to variations of human body movements and skin color when more than two blobs combined together. In this research therefore we use LBP to extract the hand(s) from face region of the image. Image is divided into nine regions to find the positions of the hand in the form of bit patterns of O's and l ' where occurrence of the hand at face region assigned otherwise 0. In this connection four common gestures are recognized such as Hands on Cheek Gesture, a lute Gesture, Face Palm Gesture and Covered Face Gesture using Artificial Neural Network. Every gesture contains nine bit patterns and feed these bits to A for recognizing particular gesture. The proposed neural network model when evaluated on our data set and found promising results with an overall accuracy of about 89.33%. Results of this research study can be very useful for many applications especially for educational institutions.