CAJ | 학술논문

为了能够有效地提取人体行为特征，提出了一种基于加速度传感器的人体行为识别系统，主要识别站立、行走、跑步、上楼、下楼这5种人体行为。该系统通过提取标准差、偏度、峰度和相关系数等统计特征来实现多层分类。实验表明，该系统可以有效地对这5种人体行为进行识别。
위료능구유효지제취인체행위특정，제출료일충기우가속도전감기적인체행위식별계통，주요식별참립、행주、포보、상루、하루저5충인체행위。해계통통과제취표준차、편도、봉도화상관계수등통계특정래실현다층분류。실험표명，해계통가이유효지대저5충인체행위진행식별。
In order to extract the characteristics of human behavior effectively, this paper proposes a system for activity recognition based on a tri-axis accelerometer. Five activity such as standing, walking, running, up stair and down stair are recognized in this system. Many Statistical characteristics are proposed including standard deviation, skewness, kurtosis and correlation coefficient for classification. Experiments show that, the system can effectively identify the five kinds of human behavior.