Gesture Recognition by Line Fitting Over Significant Pixels

Abstract

Multimedia applications have revolutionized mobile phone usage, with emphasis on interaction between user and the phone. Camera being an integral part of new generation phones can be used to felicitate this interaction using machine vision techniques. Vision algorithms have been explored to be ported on to mobile devices, however due to real-time complexity limitation these algorithms have not been largely deployed in embedded mobile applications. In this paper we propose and implement a machine vision algorithm well suited for embedded applications. The algorithm uses a predefined angular motion of human arm for camera control purpose. The algorithm being low complex is also implemented on an OMAP 2420-H4 board, which is largely used by mobile phone manufacturers.