EMG-Based Gesture Recognition System and Its Hardware Implementation

Abstract

The starting point of this method is to help disabled people control their bodies better. This paper introduces a study of gesture recognition based on electromyography (EMG). To identify the muscle control signals in gestures, we used the six lead muscle electrical sensors. Through the use of the mix of six lead mscle electrical sensor and MATLAB, the data can be read and the available dataset can be generated. We use the statistical characteristics of mean and standard deviation to determine the threshold value, so as to reduce noise and ensure acceptable data accuracy. SVM and AdaBoost algorithm are fused. In a word, the accuracy of recognition is greatly increased by using the appropriate combination classifier. In order to present better experimental results and ensure the reliability and practicability of this study, we choose a robot car as the carrier to control the movement of the robot by five gestures of four subjects.