Abstract
This paper discusses about the vision based hand gesture recognition in real time. Gesture have been the basic mode of communication since ancient times and is still used along with verbal communication. These gesture recognition systems mainly aims at empowering disabled and elderly people where they can carry out daily activities like conveying simple information or controlling machines in the form of simple hand gestures. Here a visual descriptor based system is used where skin color detection technique for hand segmentation using regular webcam is implemented. Morphological operations applied here alleviates the effects of noise. Further Haar based classifier for face detection is used to remove one of the skin colored largest contour besides hand. Then on this resultant contour, a convex hull is formed to identify the number of finger tips. As opposed to previous methods which makes use of only centroid and farthest distance calculated from centroid, fusion of convex hull algorithm and geometrical analysis proved to get a better recognition rate of 94%. These finger-tips are identified based on the geometrical analysis they make between the convexity-defect points, centroid and the hull points. Based on the number of finger-tips the gestures can be classified into various gestures. Mainly ten hand gestures based on fingertip detection has been recognized here which can be utilized by disabled people to convey these gestures in text and sound form.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call