摘要
在对基于视觉的手势识别的几个关键问题进行研究的基础上,尝试将高斯金字塔光流算法与DTW分类算法相结合,应用到动态手势识别中.实验表明,四个动态手势的平均识别率达到了94.5%.
Several key problems of vision based hand gestures recognition are discussed. And Gaussian Pyramid Optical Flow algorithm and DTW algorithm are combined to realize dynamic hand gestures recognition. The recognition rate on 4 gestures is 94.5 %.
出处
《郑州大学学报(理学版)》
CAS
2008年第3期63-66,共4页
Journal of Zhengzhou University:Natural Science Edition
基金
天津市自然科学基金资助项目
编号08JCYBJC13700
关键词
手势识别
特征选择
光流计算
DTW算法
hand gesture recognition
feature extraction
optical flow calculation
DTW algorithm