Filaments phantom data generator

In [1]:

from colorsys import hls_to_rgb

import jax as jax
import jax.numpy as jnp
import matplotlib.pyplot as plt
import numpy as np
from jax.numpy import pi

from chromatix.utils import filaments_3d

from colorsys import hls_to_rgb import jax as jax import jax.numpy as jnp import matplotlib.pyplot as plt import numpy as np from jax.numpy import pi from chromatix.utils import filaments_3d

In [2]:

# CC nadapez: from https://stackoverflow.com/a/20958684
def colorize(arr, normalize=True, gamma=0.3):
    z = arr
    r = np.abs(z)
    if normalize:
        r = r / np.max(r)

    arg = np.angle(z)

    hue = (arg + pi) / (2 * pi) + 0.5
    lightness = 1.0 - 1.0 / (1.0 + r**gamma)
    saturation = 0.8

    c = np.vectorize(hls_to_rgb)(hue, lightness, saturation)  # --> tuple
    c = np.array(c)  # -->  array of (3,n,m) shape, but need (n,m,3)
    c = c.swapaxes(0, 2)
    c = c.swapaxes(0, 1)
    return c

# CC nadapez: from https://stackoverflow.com/a/20958684 def colorize(arr, normalize=True, gamma=0.3): z = arr r = np.abs(z) if normalize: r = r / np.max(r) arg = np.angle(z) hue = (arg + pi) / (2 * pi) + 0.5 lightness = 1.0 - 1.0 / (1.0 + r**gamma) saturation = 0.8 c = np.vectorize(hls_to_rgb)(hue, lightness, saturation) # --> tuple c = np.array(c) # --> array of (3,n,m) shape, but need (n,m,3) c = c.swapaxes(0, 2) c = c.swapaxes(0, 1) return c

Generate a 3D sample with thin filaments of varying length in random orientations¶

In [3]:

arr = filaments_3d((128, 128, 128), 1, radius=0.9, rand_offset=0.2, num_filaments=50)
arr_p = jnp.sum(arr, axis=1)
plt.imshow(colorize(1j * arr_p))

arr = filaments_3d((128, 128, 128), 1, radius=0.9, rand_offset=0.2, num_filaments=50) arr_p = jnp.sum(arr, axis=1) plt.imshow(colorize(1j * arr_p))

Out[3]:

<matplotlib.image.AxesImage at 0x7f82d05d02f0>

In [4]:

z = 84
plt.imshow(colorize(1j * arr[:, z, :] * (arr[:, z, :] < 15)))

z = 84 plt.imshow(colorize(1j * arr[:, z, :] * (arr[:, z, :] < 15)))

Out[4]:

<matplotlib.image.AxesImage at 0x7f82d02b0190>

In [ ]: