基于Leap Motion的动态手势识别研究

Research on Dynamic Gesture Recognition Based on Leap Motion

基于视觉的手势识别是实现新型人机交互的一项关键技术,有其现实的研究意义。针对动态手势识别过程,开发了一个用于捕获动态手势的Gestures Visualizer系统,利用Leap Motion控制器采集数据信息和隐马尔科夫模型对手势模型进行多次训练和识别。该手势识别过程实现了动态手势数据的录制,通过对录制好的手势序列进行学习获得每个手势的模型,然后通过实验选取最优的特征集和参数,最后进行分类识别。实验结果表明,所提出的动态手势识别方法具有良好的识别率。

Gesture recognition based on vision is a key technology to realize new human-computer interaction,and it has realistic research significance.A dynamic Gestures Visualizer system for capturing dynamic gestures has been developed for the dynamic gesture recognition process,using the Leap Motion controller to collect the data information and HMM on the gesture model for several training and identification.The gesture recognition process enables the recording of dynamic gesture data,through the recognition of the recorded gesture sequence to obtain a model of each gesture.And then the optimal feature set and parameters are selected,through the experiments and finally the classification and identification is identified.Experimental results show this method has a good recognition rate in dynamic gesture recognition.