一种基于手势的虚拟场景交互方法

Method on Hand Gesture-based Interaction in Virtual Scene

手势作为一种自然的交互方式,不但方便了人与计算机的交互,而且也是对传统鼠标和键盘交互方式的一种有益补充。提出一种基于手势的虚拟环境交互方法,首先对视频采集到的手势图像进行预处理,获得手势二值图像,根据二值图像计算Zernike归一化矩,以获得手势的特征信息;然后利用BP神经网络对手势进行识别;最后根据识别的结果,驱动视点运动,实现虚拟场景的漫游。实验结果表明,该方法具有识别率高和稳定性好的优点。

Hand-gesture is a natural HCI manner.It not only facilitates the interaction between people and computers,but also provides a useful supplement on the traditional interaction as mouse and keyboard.A method on hand gesture-based interaction was proposed in virtual scene.Firstly,the binary image of hand gesture was got after image pre-processing,then zernike normalized moments was calculated to get features of hand-gesture.Finally,BP neural network was applied to hand-gesture recognition.It is successful to apply the method to walkthrough in virtual scene.The experiment results indicate that the method is high in recognition rate and stability.