Hand Grasp Recognition and Classification of Prehensile Surface EMG Signals

Abstract

Myoelectric control signals are commonly used as a convenient solution of prosthesis control for the disabled persons or amputees. Surface EMG signal being noninvasive is easy to acquire and is commonly used in prosthetic devices. Myoelectric control system is the fundamental component of modern prostheses, which uses the myoelectric signals from an individual’s muscles to control the prosthesis movements. In this paper data collected from several subjects using four surface electrodes has been used for grasp recognition. Six different (four types) grasps were used for pattern recognition or classification. Feature sets were extracted in temporal and spectral domain. The error rate was calculated for each of the wavelets using both the classifiers. The WPT feature gave the error rate equal to <5% for sym5 using MkNN classifier. The dataset being small, hence leave-one-out cross validation method was employed for testing. The focus of this work is to optimize the configuration of the classification scheme. The MkNN classifier demonstrated very good classification accuracy of >95% resulted into a robust method of grasp recognition with low computational load.

Keywords: Myoelectric signal (MES), discriminant locality preserving projections (DLPP), modified k-nearest neighbor classifier (MkNN), pattern recognition, sparse principal component analysis (SPCA)

Cite this Article

Anjana Goen, Tiwari DC. Hand Grasp Recognition and Classification of Prehensile Surface EMG Signals. Current Trends in Signal Processing. 2015; 5(3): 29–38p.