Onset detection of action surface electromyography (sEMG) signals is important since it has great impact on the accuracy of subsequent sEMG analysis such as in prosthetic control

human computer interaction and clinical diagnosis and assessment

etc.This study presents a segment onset detection method based on sample entropy.Firstly

sEMG signals are framed by a fixed-length sliding window and the sample entropy of each frame is calculated.Afterwards

adaptive threshold is set to determine the starting point.Experimental results demonstrate the feasibility of sample entropy to characterize the switching property of action sEMG signals.Meanwhile

compared to moving average and Teager-Kaiser energy operator

the proposed method has advantages in better anti-jamming ability

not only in suppressing short muscular contraction relaxation intervals

but also in suppressing involuntary background spikes.