基于otsu阈值分割和均值漂移的手势跟踪算法  

Gesture tracking algorithm based on Otsu threshold segmentation and mean shift

作　　者：赵永辉[1] 邓艳 刘立伟 徐小鹏 冯文凤 

机构地区：[1]东北林业大学机电工程学院,黑龙江哈尔滨150040

出　　处：《信息技术与信息化》2015年第11期88-92,共5页Information Technology and Informatization

摘　　要：手势跟踪是计算机视觉中,非接触式人机交互的关键技术。在复杂的环境中,受光照、旋转、形变等因素的影响,传统的均值漂移跟踪算法难以对手势进行准确的跟踪。为了解决传统均值漂移跟踪算法缺乏有效的目标更新策略,不能自动调整跟踪窗口的大小的问题,提出了基于otsu的阈值分割的均值漂移手势跟踪算法,通过otsu的阈值分割出手势目标,并实时更新手势ROI区域,通过手势ROT区域的面积确定跟踪窗口的大小,实验结果表明该算法能够对手进行实时、稳定地准确跟踪。Gesture tracking is a key technology in computer vision, and it is the key technology of non-contact human computer interaction. In a complex environment, the traditional mean shift tracking algorithm is difficult to accurately track the gestures. In order to solve the problem that the traditional mean shift tracking algorithm can not adjust the size of the tracking window automatically, the mean shift tracking algorithm based on Otsu is proposed. The algorithm can be used to determine the size of tracking window by Otsu. The results show that the algorithm can accurately track the real-time and stable ROI.

关 键 词：otsu阈值分割 均值漂移 手势跟踪 ROI区域 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]