Real Time Hand Gesture Recognition

Abstract

The automatic recognition of gestures enriches Human – Computer Interaction by offering a natural intuitive method of data input. Compared with the traditional interaction approaches, such as keyboard, mouse, pen etc. Vision based hand interaction is more natural and efficient. As this methodology is aided with cameras and computer vision techniques. And besides normal use for handling objects and manipulating tools, the hand of human being can be used as the mean of communication here. This paper presents a new method or technique for a real time static hand gesture recognition on plain uniform background, for the HCI, based on pattern recognition using K-nearest Neighbor (KNN). Major consideration here is the stable lighting condition. Here work is done to recognize 10 gestures from ASL. Concept of connected components or objects is used here for hand segmentation, canny edge detector is used for pattern creation and classification is done by using K-nearest Neighbor algorithm. This algorithm gives 84% accuracy with recognition time of 0.3s. Accuracy decreases as distance of hand from camera increases above 1.5feets. The gestures recognized here can be used further to develop an application of controlling the devices which will be very much helpful to the physically disabled and older persons who can’t move easily from one place to another.