If you have Impulse-Responce system in binary mode: either "switch-on"-regime or "switch-off"-regime - it's enough to define only these 2 regimes in filter. If you want to make them big (as you mentioned in OP) simply do:

import numpy as np from scipy import signal import matplotlib.pyplot as plt from scipy.ndimage import gaussian_filter

original = np.repeat([0., 1., 0., 0., 0., 1.], 100) impulse_response_filterKernel = [20, 1] x= np.arange(len(original))

filtered = signal.convolve(impulse_response_filterKernel, original) recovered, remainder = signal.deconvolve(filtered, impulse_response_filterKernel) print(recovered)

fig, ax = plt.subplots(1,1) ax.plot(x, original, label="original", lw=7, alpha=0.2) ax.plot(x, filtered[:len(filtered)-1], label="filtered", lw=3) ax.plot(x, recovered, label="recovered", color='black', lw=1, ) plt.legend() plt.show()