摘要
介绍了一种在一般复杂背景条件下将手势与背景分离的新方法.该方法首先在一般背景下对获得的手势图片进行卷积滤波处理,验证了肤色的聚类特性,然后综合利用手势的肤色特征和其特有的几何特征,成功地将手势与背景分离,最后通过不同背景、不同手势提取实验验证了该方法的有效性.将该手势分割方法用于8种手形手势识别实验,静态手势识别率能达到99%.
A new method which can segment gesture from general and complex backgrounds is introduced.First,this method deals the gesture picture which is get in the general background by convolute filter,and verifies clustering character of complexion,then by the complexion and geometry characters of the gestures,this method segments gestures from the backgrounds successfully.At last,by experiments of different backgrounds and eight different gestures,the validity of the method is verified,static state identify rate i...
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第S1期186-188,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家高技术研究发展计划资助项目(2007AA04Z221)
长江学者团队计划项目资助(IRT0423)
关键词
图像处理
手势
识别
分类
特征
复杂背景
image processing
gesture
recognition
classification
character
complex background
