High Density Surface Electromyography Acquisition System based on Active Electrodes and Electromyography Gesture Recognition

At present, passive electrodes have been widely used in the acquisition of weak bioelectric signals such as sEMG signals. However, due to the influence of other unfavorable factors such as power frequency noise, the collected weak biological signals have a low signal-to-noise ratio. As this power frequency noise is mixed in with useful weak signals, it greatly reduces the quality of the original signal. This poor quality raw signal affects the accuracy of extracting signal feature values and greatly increases the difficulty of preprocessing the signal. Analyzing the characteristic values of electromyographic signals and preprocessing the data are crucial for the application of electromyographic signals in gesture recognition and other fields. This article designs a 16 channel active electromyography acquisition electrode array based on FPC technology, which combines an amplification circuit with a passive electrode array. Local amplification of sEMG signals can be achieved to improve signal-to-noise ratio, which can increase by 66.67% compared to passive electrodes. At the same time, we extracted the feature values of different gestures from the electromyographic signals collected by active and passive electrodes, and classified them based on the convolutional neural network algorithm platform. Compared with the data collected by passive electrodes, the accuracy of gesture classification has been improved by 1.7%.