基于BP神经网络的手势动作表面肌电信号的模式识别

Pattern Recognition of Gesture′s Surface Myoelectrogram Signal based on BP Neural Networks

手势语言在日常生活中有着广泛的应用,本研究利用手势动作时从前臂4块肌肉上获取的4路表面肌电(SEMG)信号,经特征提取并采用BP神经网络,对8种手势动作模式进行了识别。鉴于BP网络具有较强的模式分类能力,而特征提取(幅度绝对值均值、AR模型系数、过零率)又利用了多路肌电信号的信息,实验结果取得了较高的识别正确率,表明所采用的方法是有效的。

Sign language is widely used in our daily life. In this paper, some features are extracted, using surface myoelectrogram （SEMG） signals, which were generated on four muscles of forearm when gesture actions happened. Owing to stronger classification ability of BP networks and better separability of feature vectors（which include mean absolute value, AR model parameters, and zero-crossing rate） extracted from multichannel SEMG signals, the higher accuracy was obtained in the experiments. It shows that the method is efficient.