基于穿戴式加速度传感器的手势识别系统

摘要

采用ez430-chronos智能手表作为开发平台，利用其内置功能，读取手部运动产生的加速度数据，并将其传送至处理器。对目标手势采集一定量的数据，同时采集同样量的负样本—人手的自然动作，比较二者不同之处，确定目标手势的特征，并利用该特征识别目标手势。实验结果表明：该系统对目标手势有较高识别率，而对负样本有较高的误识别率，但可以通过设定更详细的识别特征进行改善。%Theez430-chronosofTexas Instrumentisused for theez430-chronos has acceleration sensor and RF access point itself.The data of gesture canbesimply read and send to the computer.Anumber of data made from both the gestureswhich makes sense and justisdone naturallycan becollected.The difference between those gesturescan befoundout,and the gestures we need and associate them with some commandscan berecognized.It is proved that this system can recognize aim gesturewithhigher accuracy,the gestureswithoutrelated causesmore mistakes.The system canbeimprovedby changing the way in recognition.