摘要
提出一种提取手势轮廓曲线关键点对手势识别的算法。手势字母图像经过二值化处理后,提取其轮廓。将图像的轮廓看成一条曲线,采用HDC(HierarchicalDiscreteCorrelation)方法用一个内核对曲线进行多次平滑,得到曲线的尺度空间,再通过跟踪曲线在尺度空间中的运动找出手势轮廓的关键点,最后通过最小距离法对手势进行识别。
This paper presents a gesture recognition algorithm. Regarding the contour of letter gesture as a curve,we create a scale space of the curve by the application of the hierarchical discrete correlation. We then propose a new inter-scale method for feature detection based on the motion of a curve through scale space. Finally,we recognize gesture patterns be means of minimal distance of feature pixels.
出处
《安阳工学院学报》
2007年第4期69-72,共4页
Journal of Anyang Institute of Technology
关键词
关键点
字母手势
层次离散相关
最小距离
feature pixels
letter gesture
hierarchical discrete correlation
minimal distance
