Developing Human-Computer Interaction Based a Multiple-angle Hand Gesture Recognition System

Poon, Wai Hong (2012) Developing Human-Computer Interaction Based a Multiple-angle Hand Gesture Recognition System. Final Year Project (Other), Tunku Abdul Rahman University College.

Text PoonWaiHong.zip Restricted to Registered users only Download (1MB)

Abstract

This thesis study presents a gesture recognition system, which replace input devices like mouse with static and keyboard and dynamic hand gestures, for human computer interaction applications. However, there are still certain limitations and weaknesses of such systems in literature. Most applications require different parameters and constraints like having distinct lightning condition, need lots of training data, usage of a specific camera, and hand recognition by using a multi-colored glove. The system mentioned in this study eliminates all these restrictions are provides and adaptive, effort free environment to the user. Study starts with an analysis of different color space performances over skin color extraction. The analysis is independent of the working system and just performed to attain valuable information about the color space. The working system is based on two steps, namely hand detection and hand gesture recognition. In the hand detection process, many types of color space will be studied and used to threshold the coarse skin pixels in the image. Then an adaptive skin locus where differentiate boundaries are estimated from coarse skin region pixels, the segmentation of the distinct skin color in the image for the current conditions. Make use of the face detection method to apply on gesture recognition system. Gesture of the hand is recognized by centroidal profile method which is applied around the detected hand. A media player, television, a boxing game and a 3D game which are controlled remotely by using static and dynamic hand gestures were developed as human machine interface applications by using the theoretical background of this study. In experiment, the performance of the system is with the correct recognition rate of 90% was acquired with nearly real time computation.