基于多传感技术相融合的手势交互系统研究

A study of gesture interactive system based on multi-sensor technology integration

针对传统基于计算机视觉技术进行手势识别过程中存在极易受环境光照及遮挡等影响导致识别准确率和效率不高的问题,本文提出了基于MEMS传感器信号和视觉信号相融合的手势交互方法。该方法通过分别获取和处理视觉信号和MEMS传感器信号,独立计算出手势特征,最终根据不同的应用场景动态调整各种手势特征在融合模型中的权值,使其相互补充,相互修正,最终实现了根据手势特征实现人机交互的系统。实验表明,该方法的准确性和效率都很高,能够达到人机交互实时性及准确性的要求。

For the problem that gesture recognition based on the traditional computer vision technology is vulnerable to influence of ambient light and occlusion and recognition accuracy and efficiency is not high, this paper proposes a method of gesture interaction based on the integration of MEMS sensor signals and visual signals. In this method, get and process visual signal and MEMS sensor signals respectively and independently calculate gesture characteristics, and ultimately dynamically adjust feature weights based on different scenarios gestures in the fusion model to complement each other, mutual correction. Finally human-machine interaction is achieved based on gesture characteristics. Experimental results show that the accuracy and efficiency are high, and it can achieve real-time interactive and accuracy requirements.