spkit.dominent_freq

spkit.dominent_freq(X, fs, method='welch', window='hann', exclude_lower_fr=None, refine_peak=False, nfft=None, nperseg=None, return_spectrum=False, **kwargs)

Dominent Frequency Analysis

Dominent Frequency Analysis is one of the approach used to discern the different physiology from a biomedical signals [1].

This function compute dominent frequency as the maximum peak in the spectrum. The location of the peak is returned in Hz

The value of dominent frquency depends on the computing method for spectrum. Using Welch method is prefered as it computes Other posible approaches are ‘FFT’ and ‘Periodogram’

Computed peak can further be refined using parobolic interpolation.

Parameters:
X: 1d-array 2d-array

  • input signal, for 2d, channel axis is 1, e.g., (n,ch)

  • single channel or multichannel signal

fs: int,

  • sampling frequency

method: str, {‘welch’,’fft’,None}, deafult=’welch’

  • method to compute spectrum

  • welch method is prefered, (see example), uses scipy.signal.welch

  • if None or other than ‘welch’ and ‘fft’, periodogram is used to compute spectum ( scipy.signal.peridogram)

  • if ‘fft’ or ‘FFT’, dft_analysis is used.

window: str, default=’hann’

  • windowing function

exclude_lower_fr: None or scalar

  • if not None, any peak before exclude_lower_fr is excluded

  • useful to avoid dc component or any known low-frequency component

  • example: exclude_lower_fr=2, will exclude all the peaks before 2Hz

refine_peak: bool, default=False

  • if True, peak is refined using parobolic interpolation

return_spectrum: bool, default=False

  • if True, Magnitude spectrum and frequency values are returned along with dominent frequency

  • useful to analyse

(nfft,nperseg): parameters for ‘welch’ and ‘periodogram’ method
**kwargs:
  • any aaditional keywords to pass to scipy.signal.welch or scipy.signal.periodogram

    method.

Returns:
DF: scalar or 1d-array

  • scalar of X is 1d

  • array of scalers of length ch, if X is 2d

(mX, Frq)Spectrum, if return_spectrum is True

See also

clean_phase, phase_map, phase_map_reconstruction, dominent_freq_win, amplitude_equalizer

Notes

Refining peak achive better localization of dominence in spectrum. See Example.

References

Examples

#sp.dominent_freq
import numpy as np
import matplotlib.pyplot as plt
import spkit as sp
import scipy
x,fs = sp.data.optical_sample(sample=1)
x = x[:int(5*fs)]
x = sp.filterDC_sGolay(x,window_length=fs//2+1)
t = np.arange(len(x))/fs
dfq = sp.dominent_freq(x,fs,method='welch',refine_peak=True)
print('Dominent Frequency: ',dfq,'Hz')
dfq1, (mX1, frq1) = sp.dominent_freq(x,fs,method='welch',nfft=512,return_spectrum=True)
dfq2, (mX2, frq2) = sp.dominent_freq(x,fs,method='welch',nfft=2048,return_spectrum=True)
dfq3, (mX3, frq3) = sp.dominent_freq(x,fs,method='welch',nfft=512,return_spectrum=True,refine_peak=True)
dfq4, (mX4, frq4) = sp.dominent_freq(x,fs,method='welch',nfft=2048,return_spectrum=True,refine_peak=True)

plt.figure(figsize=(10,4))
plt.subplot(121)
plt.plot(frq1, 20*np.log10(mX1), label='nfft=512')
plt.plot(frq2, 20*np.log10(mX2), label='nfft=2048')
plt.axvline(dfq1,color='C0',ls='--',lw=1,label=f'df={dfq1:.3f} Hz')
plt.axvline(dfq2,color='C1',ls='--',lw=1,label=f'df={dfq2:.3f} Hz')
plt.xlim([0,20])
plt.ylim([-100, None])
plt.grid(alpha=0.1)
plt.ylabel('Spectra (dB)')
plt.xlabel('Frquency (Hz)')
plt.title('Dominent Frq.')
plt.legend()
plt.subplot(122)
plt.plot(frq3, 20*np.log10(mX3), label='nfft=512')
plt.plot(frq4, 20*np.log10(mX4), label='nfft=2048')
plt.axvline(dfq3,color='C0',ls='--',lw=1,label=f'df={dfq3:.3f} Hz')
plt.axvline(dfq4,color='C1',ls='--',lw=1,label=f'df={dfq4:.3f} Hz')
plt.xlim([0,20])
plt.ylim([-100, None])
plt.grid(alpha=0.1)
plt.xlabel('Frquency (Hz)')
plt.title('Dominent Frq. with refinement')
plt.legend()
plt.show()
../../_images/spkit-dominent_freq-1.png