中文摘要：

传统的手势指令识别在缩放、旋转、平移等转换过程中,一直存在手势图片出现形变使识别不准确。提出一种新的移动终端交互设计中手势指令识别改进方法。将质心距离函数看作全局特征对二维空间下的手势运动轨迹特征提取,对其进行改进,获取三维空间下手势轨迹特征,増加移动终端交互设计中第三维的手势轨迹信息,得到手势轨迹特征向量。将手势轨迹特征向量作为输入,通过随机森林方法对手势指令识别分类器进行训练,利用训练完成的分类器实现移动终端交互设计中手势指令的识别。实验结果表明,所提方法不仅能够保证移动终端交互设计中手势指令识别的实时性,而且识别精度高,认证效果好。

英文摘要：

In view of the traditional gesture command recognition has been images in scaling, rotation, translation, such as deformation appeared in the process of transformation, to identify the problem of inaccurate, a new gesture command recognition improvement methods was put forward in the design of mobile terminal interaction. The centroid distance function as a gesture of global features of two-dimensional space trajectory characteristics were extracted, to improve, get the gesture trajectory characteristic under the three dimensional space, rights and the third dimension in the design of mobile terminal interaction gesture trajectory information, get gesture trajectory characteristic vector. Will gesture trajectory characteristic vector as input, instruction by means of random forest gestures recognition classifier was trained, using classifier training completed in the design of mobile terminal interaction gestures recognition of instruction. The experimental results show that the proposed method can not only ensure mobile terminal interaction design real-time gesture recognition instruction, and the identification accuracy is high, effect is good.