高斯径向基函数重构特征对表面肌电信号识别

针对在不同动作模式下对表面肌电信号提取的特征信息总是有较大差异，而相同动作模式下提取的特征信息较为接近这一特点，提出了高斯径向基函数重构算法对肌电信号进行识别。该算法在对表面肌电信号提取特征信息后，用高斯径向基函数对特征矢量进行重构，使得重构的特征矢量的空间分布存在很大差异而直接进行识别。用该重构算法对提取的AR系数重构，然后进行识别，平均识别率为97.2%；对小波系数重构，平均识别率为99%。

Recognition of sEMG based on reconstructed feature by Gaussian radial basis function

Since the extracted feature information of sEMG is always quite different in different action modes and is close to each other in the same action modes, it is very useful in an algorithm based on Gaussian Radial Basis Function（RBF） to reconstruct the extracted feature information. The algorithm extracts feature information from sEMG. The Gaussian RBF is used to reconstruct the feature information. According to the different distribution of the reconstructed feature information, the sEMG can be identified directly by the reconstructed feature information. Using the Autoregressive（AR） coefficients as the extracted feature information, the average recognition rate of the reconstruction algorithm is 97.2%; using the wavelet coefficients, the average recognition rate is 99%.