squiggle_processing_utils.py

import scipy.signal as scisignal
import numpy as np
from dtw import dtw
from dtw import warp as dtw_warp

def de_noise_signal(signal):
    # TODO track down this justification
    # Bessel filter order and Wn chosen based on how good some chart looked to me at some point
    fs = 3012  # Sampling frequency in Hz
    cutoff_freq = 100  # Cutoff frequencies in Hz
    nyquist = 0.5 * fs
    Wn = cutoff_freq / nyquist
    order = 8

    b, a = scisignal.bessel(order, Wn, 'low')
    return scisignal.filtfilt(b, a, signal)

def compute_dtw_distance(signal1, signal2):
    alignment = dtw(signal1, signal2, step_pattern="symmetric1", distance_only=True)
    return alignment.distance

def compute_dtw_warp(signal1, signal2):
    alignment = dtw(signal1, signal2, step_pattern="symmetric1", keep_internals=True)
    return dtw_warp(alignment, index_reference=False)

def scale_and_shift(signal, scale, shift):
    return signal * scale + shift

def rescale_to_range(signal, new_min, new_max):
    scale, shift = get_rescale_factors(signal, new_min, new_max)
    return scale_and_shift(signal, scale, shift)

def get_rescale_factors(signal, new_min, new_max):
    signal = np.array(signal)
    orig_min = np.min(signal)
    orig_max = np.max(signal)  
    scale = (new_max - new_min) / (orig_max - orig_min)
    shift = new_min - orig_min * scale
    return scale, shift