计算机工程与应用 ›› 2016, Vol. 52 ›› Issue (9): 202-207.

• 图形图像处理 • 上一篇    下一篇

基于时空上下文的手势跟踪与识别

王彩红1,2,沈燕飞2,3,王  毅1,朱珍民2,3   

  1. 1.湘潭大学 信息工程学院,湖南 湘潭 411105
    2.中国科学院 计算技术研究所,北京 100190
    3.移动计算与新型终端北京市重点实验室,北京 100190
  • 出版日期:2016-05-01 发布日期:2016-05-16

Gesture tracking and recognition based on spatio-temporal context

WANG Caihong1,2, SHEN Yanfei2,3, WANG Yi1, ZHU Zhenmin2,3   

  1. 1.College of Information Engineering, Xiangtan University, Xiangtan, Hunan 411105, China
    2.Institute of Computing Technology, Chinese  Academy of Sciences, Beijing 100190, China
    3.Beijing Key Laboratory of Mobile Computing and Pervasive Device, Beijing 100190, China
  • Online:2016-05-01 Published:2016-05-16

摘要: 针对复杂背景下手势运动过程中出现的手势形态变化、遮挡、光照变化等问题,提出了一种基于时空上下文的手势跟踪与识别方法。使用机器学习方法离线训练手势样本分类器,实现对手势的检测和定位;利用时空上下文跟踪算法对动态手势进行跟踪,同时为了避免跟踪过程中出现的漂移、目标丢失等情况,使用手势检测算法对手势位置信息进行实时校准;根据手势运动轨迹对手势运动进行跟踪与识别。实验表明,提出的方法可以实现对手势运动快速、准确、连续识别,满足人机交互的要求。

关键词: 检测, 手势跟踪, 时空上下文

Abstract: A gesture tracking and recognition method based on spatio-temporal context is proposed to solve the problems such as gesture posture variation, occlusion, illumination changes under the complex background. Firstly, in order to get the location information of gesture, machine learning is used for off-line training a gesture classifier; then spatio-temporal context algorithm is used to track dynamic gesture. At the same time, the?gesture?detection algorithm is used for?real-time?calibration?to the gesture?position information in order to avoid?drift, target?lost in the tracking process. Finally, according to the gesture?trajectory?tracking, hand movements can be identified. Experimental results show that the proposed algorithm can achieve rapid, accurate and continuous recognition of the gesture motion and meets the requirements of the human-computer interaction.

Key words: detection, gesture tracking, spatio-temporal context