| Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 99 === This thesis presents two accelerometer-based applications: a fall detection system and a gesture recognition system. Recently, there are several approaches to fall detection. Each approach has its own advantages, disadvantages, and limitations. The first objective of this thesis is not only to detect falls but also to identify the directions of falls based on a a tri-axis accelerometer. The proposed fall detection system incorporated with a ZigBee-based location system can quickly locate the position where a fall happens such that a quick and effective response can be issued. In order to increase the error tolerance and decrease the miss-classification of the activities of daily living, the fall detection system adopts a fuzzy system to implement the decision core module of the fall detection system. For the time being, the fall detection system can identify four directions and the correct recognition rate was about 95%. In addition, we have already implemented the fall detection system in a microprocessor to increase its applicability. The second objective of the thesis is to develop a hand gesture recognition system. An accelerometer is adopted to record a user’s hand trajectories. The trajectory data is transmitted wirelessly via an RF module to a computer. Then the dynamic time warping (DTW) algorithm is adopted to classify six different hand trajectories. Simulation results show that the recognition rate could achieve 92.2% correct. Finally, the proposed hand gesture recognition system was adopted for navigating a car-robot.
|
|---|
