A Histogram Method for Detecting Useful Surface Electromyogram Signals

摘要

SEMG (Surface Electromyogram) is a common noninvasive method of measuring the electrical activity in muscle tissue. The SEMG signal is used in various applications, from clinical research to application prostheses and human-computer interfaces. The SEMG signal is a small amplitude signal and strongly affected by noise. The aim of this study is to design and implement a hardware method of signal extraction in the time domain, without affecting useful SEMG amplitude and phase characteristics, so with minimal distortion. This approach allows isolating a segment in the time domain of the SEMG signal, based on the statistical histogram method. The selection criterion relies on choosing a signal area with a maximum number of samples that exceed the threshold in the required histogram. The imposed threshold depends on the signal-to-noise ratio of the acquired SEMG signal. We have designed and built equipment for SEMG signal amplification, conversion, and processing. A SEMG signal selection algorithm has been hardware implemented using a Xilinx XUPV5 development board with Virtex-5 FPGA. Experimental results suggest that this algorithm is simple, fast, and easy to implement as a hardware solution. The algorithm reduces the number of SEMG signal samples processed and thus the computation time.