Hand gesture recognition using sparse autoencoder-based deep neural network based on electromyography measurements

Hand gesture recognition has recently grown as a powerful technical means in human-machine interaction field for control the appliances such as in home automation. However, the accuracy recognition of diverse hand gestures is still in the early stage for real-world application. In this paper, we present a new gesture recognition framework which is capable of classifying ten different hand gestures based on the input signals from surface electromyography (sEMG) sensors. The multi-channel signals of a hand motion are simultaneously captured and transmitted to a PC via Bluetooth wireless protocol. The proposed recognition framework composes of three main steps: gesture sequence segmentation, feature extraction by sparse autoencoder, and deep neural network (DNN) based classification. The advantage of the proposed approach is the automated abstract feature extraction based on sparse autoencoder method. Combined with the DNN classification technique, we could achieve a better recognition performance tested on the dataset consisting of ten types of hand gestures compared with other classification methods.