ABSTRACT
ABSTRACT: This paper presents a surface Electromyographic (sEMG) signal based hand grasp recognition technique utilizing Empirical Mode Decomposition (EMD) and Differential Evolution based Feature Selection (DEFS). A series of features were derived from both the raw signal and its corresponding Intrinsic Mode Functions (IMF’s), obtained by performing EMD. Differential Evolution (DE) is a relatively new soft computing technique with wide range of applications. Being a promising stochastic population based optimization method, a feature selection framework using DE is utilized in this research to identify the optimum feature subset. sEMG signals recorded from eleven healthy subjects are used for this study. The proposed method is further validated using other popular feature extraction techniques and different pattern recognition algorithms. Outcome of our research shows that the methodology of using EMD with DEFS can achieve significant improvement in the overall recognition rate.
