Abstract: As effective stage extraction of surface electromyography (sEMG) is the premise of signals analysis and processing,an improved method has been proposed to realize sEMG segmented automatically which could lay a foundation for full automatic analysis of rehabilitation robot．In this paper the value of window energy has been taken as the judgment standard for the start-stop point of muscle action,and the formula of initial threshold has been given．The wavelet transform technique has been applied to filter the non-action signals,and the segmentation value has been adjusted automatically according to the characteristics of the split point．The experiments results show that this method can segment sEMG signals automatically,without regarding as the artificial factors of conner,without initialization,and the results are more accurate and the precision is higher.

Key words: sEMG,Wavelet transform,Signal segmentation,Rehabilitation robot