基于雷达技术的手势识别

Gesture Recognition Based on Radar Technology

本文提出了一种基于雷达技术的手势识别方法,利用手势动作中产生的多普勒频率信息作为手势的特征信息,采用了雷达技术中常用的线性调频、压缩处理、动目标显示作为手势信号处理算法,有效地获取不同手势信号的特征信息再予以成像,而后针对图像采用支持向量机算法进行分类和识别。上述理论基础之上,本文设计了一个拥有七种不同手势的手势识别系统进行实验,最终手势识别正确率达到了92.9%,成功验证了本文手势识别方法是准确有效的。

This article presents a gesture recognition method which is based on radar technology, consid- ering the Doppler frequency information caused by hand motion as feature information of gestures and tak- ing advantages of the technology commonly used in radar system, such as chirp signal, signal compression, MTI. It is a signal processing algorithm which has an efficient access to getting characteristic information of different gestures and making them to be imaged. And then for the images, support vector machine classification algorithm will be a reference to image recognition. In the basis of the theory above, we design a gesture recognition system with seven different kinds of gesture to experiment. The final gesture recognition accuracy rate reached 92.9%, successfully demonstrated that the method above was accurate and efficient.