FIRFilter

class eyefeatures.preprocessing.smoothing.FIRFilter(x, y, t=None, pk=None, numtaps=81, fs=250, cutoff=100, pass_zero=False, mode='valid', **fir_kw)

Bases: BaseSmoothingPreprocessor

FIR filter. Convolution with FIR kernel along ‘x’ and ‘y’. kwargs are passed to scipy.signal.firwin to determine the kernel.

Parameters:

• mode (Literal['valid', 'full', 'same']) – parameter of scipy.signal.fftconvolve.

• x (str)

• y (str)

• t (str)

• pk (list[str])

• numtaps (int)

• fs (int)

• cutoff (float | tuple[float, ...])

• pass_zero (Literal[False, True, 'bandpass', 'lowpass', 'highpass', 'bandstop'])

Notes

Default values are taken from https://arxiv.org/pdf/2303.02134.

Algorithm

Weighted sum of \(k\) previous values (convolution of signals).