HAND GESTURE RECOGNITION

Hand gestures are a primary mode of communication for speech and hearing-impaired individuals. However, the lack of a universal mechanism in place to interpret these gestures automatically and express them as comprehensible speech remains one of the major barriers to communication. This work proposes an architecture of hand gesture recognition and speech synthesis, where real-time hand gestures of a speaker captured through a camera are conveyed into text and synthesised speech output after processing through contour-based image analysis. Contour-based image analysis uses image preprocessing, extraction of contours, convex hull detection, and defect analysis to effectively identify the gesture corresponding to a predefined sign language symbol. Once correctly identified, the gesture is translated into speech with the help of an integrated text-to-speech mechanism that will allow the establishment of natural and fluent communication between impaired and non-impaired individuals. This proposed approach is low-cost, eliminates the use of extra sensors or any wearable gadget, and thus presents a very user-friendly platform for everyday applications. Experimental results are encouraging: fair accuracy and good responsiveness were achieved, showing the potential of this system as an aid in communication in visually, vocally, and hearing-impaired communities.