Hand Gesture User Interface for Smart Devices Based on Mems Sensors

This paper presents a MEMS accelerometer based handheld embedded device and its associated dynamic time warping (DTW) based algorithm for recognizing hand writing gestures. Device can be used to draw gestures for English lower case letters, Arabic numerals in their own style and speed. Furthermore the proposed device can detect different direction movement such as left, right, up, down, push and pull for employing video game and mobile game control. The accelerometer signals are processed within the device itself and result of recognition is transmitted via Bluetooth to smart device such as computer, smart phone, tablets etc. Furthermore we have introduced a novel method of fast and memory efficient implementation of DTW in embedded domain. Also we have developed an effective motion detection method based on adaptive mean value calculation of accelerometer signals also we introduced an effective template selection criterion. The effectiveness of proposed system and associated algorithm were validated through various experiments.