实时手势加速度动作分割与识别研究  

Accelerometer Based Real-time Gesture Segmentation and Recognition

作　　者：刘蓉[1] 刘明[2] 

机构地区：[1]华中师范大学物理科学与技术学院,武汉430079 [2]华中师范大学计算机科学系,武汉430079

出　　处：《小型微型计算机系统》2012年第7期1620-1624,共5页Journal of Chinese Computer Systems

基　　金：华中师大中央高校科研业务费项目(CCNU10A02008)资助;教育部人文社会科学研究规划项目(10YJA870026)资助

摘　　要：实时手势动作分割与识别是基于惯性传感器手势交互的重要研究内容.采用佩戴在手腕的单个加速度传感器获取手势加速度信号,提出一种实时手势加速度动作分割和识别方案.首先采用基于阈值的动作分割算法实时切分连续手势,通过聚类算法提取手势动作的关键特征,然后构造离散隐马尔可夫模型实现手势识别.实验结果表明,本文采用的手势动作切分算法能自动提取有效手势信号,关键特征选择不仅降低了隐马尔可夫模型的复杂度,而且提高了识别率.Real-time gesture segmentation and recognition are very important task in inertial sensor-based HCI（human-computer interaction） research.This paper uses a three-axial accelerometer to capture the acceleration of dominant wrist,and a real-time gesture segmentation and recognition approach is proposed.It achieves auto temporal segmentation by using the gesture segmentation algorithm based on slope threshold and error threshold of fitting lines,then choice the key features by k-means clustering,and Discrete Hidden Markov Models are built for gesture recognition.Experimental results show that the proposed method can achieve real-time gesture segmentation,and the key feature selection method can not only reduce the complexity of Discrete Hidden Markov Models,but also improve the recognition performance.

关 键 词：手势识别 加速度传感器 动作分割 特征选择 隐马尔可夫模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]