`detectda` demo

In the below, test_video.pkl is the output of the “identify_polygon” python script,
which allows choose your desired polygonal subregion.

from skimage import io, filters
import detectda as dtda
import numpy as np
import pickle
import matplotlib.pyplot as plt
import cv2

"""
Here we divide each pixel by 32 (div=32) and round to the nearest integer because of various properties
of the detector which captured the video. Outside of hypothesis testing context, it is fine to set div equal to
its default value of div=1!
"""

impol = dtda.ImageSeriesPickle('detectda/tests/test_video.pkl', div=32, n_jobs=2)

Calculate persistent homology

Fit the persistence diagrams for every image in the polygonal region of the test video. We found smoothing parameter of 2 worked well for these nanoparticle videos.

impol.fit(sigma=2)

Visualize the results of the fit

See where the homology generators appears and how long they live. Below, we see the raw image with the points of the 0^th persistence diagram overlaid, after removing points with lifetimes less than thr.

impol.plot_im(73, thr=0.7, smooth=False)

png

Calculate summaries

Run these routines to get persistent entropy and the ALPS statistic

impol.get_pers_entr(neg=False)
impol.get_alps()

Plot the summaries

Plot the ALPS statistic across all the frames

plt.plot(impol.alps) #or impol.pers_entr
plt.xlabel('Frame')
plt.ylabel('ALPS statistic')

png

Hypothesis testing

Run hypothesis testing using the observed image series above, by generating 500 Monte Carlo noise images

G = open('detectda/tests/test_video_vacuum.pkl', 'rb')
tv_vacuum = pickle.load(G)['video']
impol_vac = dtda.VacuumSeries(tv_vacuum, observed_ImageSeries=impol, parametric=False, div=32)
impol_vac.fit(convert_to_int=True)
impol_vac.transform(500, "alps", conservative=True)

Plot the test results

Plot the results of the hypothesis testing

impol_vac.plot_hypo()

png

detectda demo