基于头势的人机交互系统研究  

Research on the human-computer interaction system based on head gesture

作　　者：翁磊[1] 沈天飞[1] 张贺[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072

出　　处：《电子测量技术》2013年第7期51-54,共4页Electronic Measurement Technology

摘　　要：近年来,肢体残疾人数增长迅速,为了改善他们的生活质量和生活自由度,从方便且易于残疾人使用的角度出发,提出了一种利用头势来控制轮椅的方法。首先根据肤色信息找到人脸区域,然后利用AdaBoost算法在人脸区域找到鼻子。再根据计算鼻子区域的质心在整个画面上的位置信息识别头部姿态,从而达到控制轮椅运动的目的。实验结果表明,该方法可以准确的识别头部姿态,平均识别率达到了89.55%,同时具有较快的检测速度,且操作便易。In recent years,physically disabled persons have been increased rapidly. In order to improve the quality and freedom of their lives, this paper proposes a head gesture control method on intelligent wheelchair. Firstly, detect the face area based on skin information,then find the nose from the face region by Adaboost algorithm. The head gesture is recognized on the basis of the nose cancroids~ relative position. The head gesture commands determine the movement of the wheelchair. Experiments show that this approach can recognize the head gesture correctly and the average rate of recognition is 89.55% ,at the same time it has great detection speed and easy to operate.

关 键 词：头势 智能轮椅 肤色分割 ADABOOST 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]