基于加速度特征的可拓动作识别方法

Extension Action Recognition Method Based on Acceleration Feature

针对基于图像和视频的动作识别系统具有特征采集设备复杂、视角固定和需要采集多视角图像等缺点,提出基于加速度特征的可拓动作识别方法。该方法利用物体向不同方向运动时,其关键部位点的三轴加速度具有一定区分度的特点,结合可拓识别方法,实现动作识别。在构建的手臂动作识别系统中,测得动作识别率可达94.4%。该方法可应用于智能监控、医疗电子等领域。

Extension action recognition system based on acceleration feature is proposed to get rid of the shortcomings that the features gathering equipments are complicated, the angle of view is fixed and multi-angle pictures are needed in action recognition system which bases on pictures and videos. This action recognition method makes use of the property that the accelerations of important locations are different as object moves along different directions, and combines the extension recognition method. In the paper, a simulative system was constructed to recognize three types of arm action with the method of extension pattern recognition which bases on the acceleration features, and the recognition rate was 94.4%. This method will be wildly applied to intelligent control and medical electronics.