| import numpy as np | |
| from scipy.signal import butter, lfilter, filtfilt, freqz | |
| from scipy import signal | |
| def BPfilter(x, minHz, maxHz, fs, order=6): | |
| """Band Pass filter (using BPM band)""" | |
| #nyq = fs * 0.5 | |
| #low = minHz/nyq | |
| #high = maxHz/nyq | |
| #print(low, high) | |
| #-- filter type | |
| #print('filtro=%f' % minHz) | |
| b, a = butter(order, Wn=[minHz, maxHz], fs=fs, btype='bandpass') | |
| #TODO verificare filtfilt o lfilter | |
| #y = lfilter(b, a, x) | |
| y = filtfilt(b, a, x) | |
| #w, h = freqz(b, a) | |
| #import matplotlib.pyplot as plt | |
| #fig, ax1 = plt.subplots() | |
| #ax1.set_title('Digital filter frequency response') | |
| #ax1.plot((fs * 0.5 / np.pi) * w, abs(h), 'b') | |
| #ax1.set_ylabel('Amplitude [dB]', color='b') | |
| #plt.show() | |
| return y | |
| def zeroMeanSTDnorm(x): | |
| # -- normalization along rows (1-3 channels) | |
| mx = x.mean(axis=1).reshape(-1,1) | |
| sx = x.std(axis=1).reshape(-1,1) | |
| y = (x - mx) / sx | |
| return y |