| Summary: | 碩士 === 國立中央大學 === 資訊工程學系碩士在職專班 === 98 === Gesture recognition is a popular research topic at present, its application is broad which includes robot operation, controller for video games or other household appliances and etc. The purpose of this paper is to design a Human-Machine interface by using a video camera to capture live hand gesture to achieve the goal of hand gesture recognition. This system will track the hand movement using background subtraction, morphology and CamShift algorithm to extract the hand contour. After the hand contour is obtained, the system will then identify each finger according to its unique feature and calculate the angle between fingers and mass centre of palm, using the number of fingers and its angles to determine the command given by the gesture. Finally, the command will be passed to STB through an infrared transmitter.
For result verification we have defined 8 types of hand gesture, the experiment is conducted with 7 participants aiming at 5 scenarios. The average recognition rate is 94.1%.
|
|---|
