华中科技大学学报（自然科学版）Issue(z1)：62-65,4.DOI:10.13245/j.hust.15S1015
基于深度信息的无标记手势分割与3D 定位
Marker-less gesture segmentation and 3D positioning based on depth information
摘要
Abstract
A accurate and stable method of gesture segmentation and 3D (three dimensional)positio-ning was proposed based on Kinect depth information.The gesture binary image was detected based on depth threshold in experimental space.According to behavioral characteristics of gestures,accurate gesture was segmented without marker by endpoint detection and dynamic threshold algorithm.In or-der to avoid imprecise segmentation,the result of segmentation was limited by threshed.The method of 3D positioning was proposed based on barycenter coordinate and average gray value.The experi-ments demonstrate that the method of segmentation possess better segmentation accuracy,and the 3D positioning has a better stability than skeletal tracking of Kinect SDK,and it has no singular point.关键词
无标记手势分割/3D 定位/深度信息/实验空间/动态阈值Key words
marker-less gesture segmentation/3D positioning/depth information/experimental space/dynamic threshold分类
信息技术与安全科学引用本文复制引用
林海波,周庭,张毅..基于深度信息的无标记手势分割与3D 定位[J].华中科技大学学报（自然科学版）,2015,(z1):62-65,4.基金项目
科技部国际科技合作资助项目（2010DFA12160）． （）
