A Multi-Gestures Recognition System Based on Less sEMG Sensors

With complex functions, hand is an important organ for human. Unfortunately, many people in China are suffering from hand losing. Therefore, the effective hand motion recognition system is required to help the amputees live or work normally. The surface electromyography (sEMG) signal can represent the hand motion effectively, and many studies about sEMG-based prosthetic hands have been investigated. However, some prosthetic hands use on-off switch control command, which limits the intelligence and flexibility of the prosthetic hands. Some intelligent recognition systems require too many sensors, which is unrealistic for amputees with limited residual muscles. In addition, some algorithms are too complicated, which brings difficulties for practical applications. To solve these problems, we attempted to recognize six commonly used handgestures with two-channel sensors, and the classification performance and calculation time of different algorithms are compared. Finally, we achieved the recognition accuracy of 91.93% by three time domain features and back propagation neural network (BPNN) classifier, which balances the accuracy and computation time. In future work, the proposed method will be applied to real-time prosthetic hands to improve the amputee’s quality of life.