摘要
针对静态手势,提出一种基于形状特征的手势识别方法。使用深度信息与肤色信息相结合的方法,分割出手势区域;对分割后的手势进行卷积和阈值化处理,得到手指部分的图像信息,根据手掌质心和手指质心准确判断手的方向,保证手的旋转不变性;在极坐标下,结合手指特征能够准确得到每个手指的具体角度和弯曲状态,判断出左右手。对比实验结果表明,该方法对手势的判别和左右手判别具有更高的性能、更强的鲁棒性。
For static gestures,a gesture recognition method based on shape feature was put forward.The method combining with color information and depth information was used to segment the gesture area.Through convolution and threshold,the segmented gestures were processed and the image information of finger part was got.According to the center of mass of palm and fingers,the direction of the hand was accurately judged to ensure the rotation invariance of hand.According to the feature of finger under polar coordinates,the concrete angle and bending state of each finger were got,while the right hand was judged.Results of contrast experiment show the proposed method has better performance and stronger robustness in judging hand gestures.
作者
谭台哲
韩亚伟
邵阳
TAN Tai-zhe;HAN Ya-wei;SHAO Yang(College of Computer,Guangdong University of Technology,Guangzhou 510006,China;Synergy Innovation Institute of GDUT,Heyuan 517001,China)
出处
《计算机工程与设计》
北大核心
2018年第2期511-515,共5页
Computer Engineering and Design
