import matplotlib.pyplot as plt
import neurokit2 as nk
import numpy as np
SAMPING_RATE = 2000
k = 4
n = 2
[docs]def process_one_channel(data, sampling_rate, k):
"""
:param data: raw emg data
:param sampling_rate: recorded at samples / second
:param k: filtering rate of data to samples / k / second
:return:
data_filter: Filter the original EMG signals (from the Delsys system, 2000 Hz) to the desired frequency
data_clean: Clean the raw EMG signals
data_mvc: Calculate the Maximum Voluntary Contraction (MVC) of the EMG signals
data_abs: Take the absolute value of the EMG signals
"""
data_filter = []
for i in range(0, len(data) - k + 1, k):
data_new = data[i]
data_filter.append(data_new)
data_filter = np.array(data_filter)
signals, info = nk.emg_process(data_filter, sampling_rate=sampling_rate)
signals_array = signals.values
data_clean = signals_array[:, 1]
data_mvc = signals_array[:, 2]
data_abs = [0] * len(data_clean)
for i in range(len(data_clean)):
data_abs[i] = abs(data_clean[i])
data_abs = np.array(data_abs)
return data_filter, data_clean, data_mvc, data_abs
[docs]def process_all_channels(data, n, sampling_rate, k):
"""
:param data: raw emg data
:param n: number of emg channels
:param sampling_rate: recorded at samples / second
:param k: filtering rate of data to samples / k / second
:return:
DATA_FILTER: Filter the original EMG signals (from the Delsys system, 2000 Hz) to the desired frequency
DATA_CLEAN: Clean the raw EMG signals
DATA_MVC: Calculate the Maximum Voluntary Contraction (MVC) of the EMG signals
DATA_ABS: Take the absolute value of the EMG signals
"""
DATA_FILTER = []
DATA_CLEAN = []
DATA_MVC = []
DATA_ABS = []
DATA_ACTIVITY = []
DATA_MVCSCALE = []
for j in range(n):
data_filter = []
for i in range(0, len(data[:, j]) - k + 1, k):
data_new = data[i, j]
data_filter.append(data_new)
data_filter = np.array(data_filter)
signals, info = nk.emg_process(data_filter, sampling_rate=sampling_rate)
signals_array = signals.values
data_clean = signals_array[:, 1]
data_mvc = signals_array[:, 2]
data_activity = signals_array[:, 3]
data_abs = [0] * len(data_clean)
for i in range(len(data_clean)):
data_abs[i] = abs(data_clean[i])
data_abs = np.array(data_abs)
data_mvcscale = (data_mvc - data_mvc.min()) / (data_mvc.max() - data_mvc.min())
DATA_FILTER.append(data_filter)
DATA_CLEAN.append(data_clean)
DATA_MVC.append(data_mvc)
DATA_ACTIVITY.append(data_activity)
DATA_ABS.append(data_abs)
DATA_MVCSCALE.append(data_mvcscale)
DATA_FILTER = np.transpose(np.array(DATA_FILTER), (1, 0))
DATA_CLEAN = np.transpose(np.array(DATA_CLEAN), (1, 0))
DATA_MVC = np.transpose(np.array(DATA_MVC), (1, 0))
DATA_ACTIVITY = np.transpose(np.array(DATA_ACTIVITY), (1, 0))
DATA_ABS = np.transpose(np.array(DATA_ABS), (1, 0))
DATA_MVCSCALE = np.transpose(np.array(DATA_MVCSCALE), (1, 0))
return DATA_FILTER, DATA_CLEAN, DATA_MVC, DATA_ACTIVITY, DATA_ABS, DATA_MVCSCALE
[docs]def plot_raw_and_clean(data_filter, data_clean, k):
fig, ax0 = plt.subplots(nrows=1, ncols=1, sharex=True)
ax0.set_xlabel("Time (seconds)", fontweight="bold", fontdict={'family': 'Times New Roman'}, fontsize=12)
fig.suptitle("Raw and Clean EMG Signals", fontweight="bold", fontdict={'family': 'Times New Roman'},
fontsize=16)
plt.subplots_adjust(hspace=0.2)
x_axis = np.linspace(0, data_filter.shape[0] / int(2000 / k), data_filter.shape[0])
legend_font = {"family": "Times New Roman"}
ax0.set_title("Sensor", fontdict={'family': 'Times New Roman'}, fontsize=12)
ax0.plot(x_axis, data_filter, color="#B0BEC5", label="Raw", zorder=1)
ax0.plot(
x_axis, data_clean, color="#FFC107", label="Cleaned", zorder=1, linewidth=1.5
)
ax0.legend(loc="upper right", frameon=True, prop=legend_font)
[docs]def plot_abs_and_mvc(data_abs, data_mvc, k):
fig, ax0 = plt.subplots(nrows=1, ncols=1, sharex=True)
ax0.set_xlabel("Time (seconds)", fontweight="bold", fontdict={'family': 'Times New Roman'}, fontsize=12)
fig.suptitle("Absolute Value and MVC of EMG signals", fontweight="bold",
fontdict={'family': 'Times New Roman'}, fontsize=16)
plt.subplots_adjust(hspace=0.2)
x_axis = np.linspace(0, data_abs.shape[0] / int(2000 / k), data_abs.shape[0])
legend_font = {"family": "Times New Roman"}
ax0.set_title("Sensor", fontdict={'family': 'Times New Roman'}, fontsize=12)
ax0.plot(x_axis, data_abs, color="#B0BEC5", label="ABS", zorder=1)
ax1 = ax0.twinx()
ax1.plot(
x_axis, data_mvc, color="#FA6839", label="MVC", linewidth=1.5
)
ax0.legend(loc="upper left", frameon=True, prop=legend_font)
ax1.legend(loc="upper right", frameon=True, prop=legend_font)
[docs]def plot_mvcscale_and_activity(data_mvcscale, data_activity, k):
fig, ax0 = plt.subplots(nrows=1, ncols=1, sharex=True)
ax0.set_xlabel("Time (seconds)", fontweight="bold", fontdict={'family': 'Times New Roman'}, fontsize=12)
fig.suptitle("MVC and threshold of EMG signals", fontweight="bold",
fontdict={'family': 'Times New Roman'}, fontsize=16)
plt.subplots_adjust(hspace=0.2)
x_axis = np.linspace(0, data_mvcscale.shape[0] / int(2000 / k), data_mvcscale.shape[0])
legend_font = {"family": "Times New Roman"}
ax0.set_title("Sensor", fontdict={'family': 'Times New Roman'}, fontsize=12)
ax0.plot(x_axis, data_mvcscale, color="#B0BEC5", label="MVC", zorder=1)
ax1 = ax0.twinx()
ax1.plot(
x_axis, data_activity, color="#FA6839", label="threshold", linewidth=1.5
)
ax0.legend(loc="upper left", frameon=True, prop=legend_font)
ax1.legend(loc="upper right", frameon=True, prop=legend_font)
# if __name__ == '__main__':
# emg = np.load('../../data/emg_data.npy')
# data_filter, data_clean, data_mvc, data_abs = process_all_channels(emg, n, SAMPING_RATE, k)
#
# for i in range(n):
# plot_raw_and_clean(data_filter[:, i], data_clean[:, i], k)
# plot_abs_and_mvc(data_abs[:, i], data_mvc[:, i], k)
# plt.show()
# # process single channel
# data_filter_1, data_clean_1, data_mvc_1, data_abs_1 = process(emg[:, 0], SAMPING_RATE, n)
# data_filter_2, data_clean_2, data_mvc_2, data_abs_2 = process(emg[:, 1], SAMPING_RATE, n)
