中文摘要：

虚拟现实中的交互手势包括多种不同类型，层次化建模方法避免了采用单一模型导致效率不高的问题．识别是一个由粗到精的过程，通过滑动窗技术实时提取手势的统计特征，实现手势类别的粗略划分，然后采用不同方法对各类手势进行分析．交互环境及上下文信息用以辅助手势的类别划分，提高了识别效率．最后通过虚拟家居系统对该方法进行了验证．

英文摘要：

Since interactive the problem caused by using gestures in VR systems have variety of single model. Gesture recognition is a types, hierarchical method can solve coarse-to-fine process. By means of sliding window technology, statistic features of gestures are extracted in real time and used to achieve coarse gesture taxonomy. Subsequently, different analytical methods are applied to different gesture categories to accomplish interaction tasks. Besides, environmental information and context are employed to improve gesture understanding. At last, a gesture-based virtual room system is introduced to validate the proposed method.