Somayeh,

EMG generally results from a bi-polar signal measurement of muscle activity with both positive and negative components as the activation occurs. As such, the signal is usually measured at a high frequency to pick up the signal's relative positive and negative directions. So merely downsampling or using a spline to decrease the number of samples will remove important parts of the signal.

That being said, the positive-negative parts of that signal are usually physiologically challenging, if not impossible, to interpret and what is often of more interest is the timing of increasing or decreasing muscle activation, which is correlated to the net amplitude of the signal as time progresses. Increased activation yields an increased EMG signal as more motor units are recruited via a more significant electrical signal.

Therefore, you should first use a moving root mean squared (RMS) window that takes the RMS of a certain number of samples (n) at a time and progresses to the end of the waveform. The RMS window process will then remove the positive-negative component of the signal and leave only information related to the modulating amplitude of the signal. In the more general signal processing realm, this is often referred to as "demodulation." After you have RMS windowed/demodulated your signal, you can simply use a spline filter or a standard downsampling/algorithm to get your data and your desired number of samples for your 100 data points for the complete movement phase for the individual.

Thor

Yes! The situation becomes worse when religious discrepancy is held supreme at the expense of scholarship. I once sent an article to some universities in Saudi Arabia and Bahrain with an address of a university from a country with different religious sect. To my surprise, the editors comments on the article were positive and despite that, they expressed reservation from publishing it.