1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
|
# Copyright (c) DataLab Platform Developers, BSD 3-Clause license, see LICENSE file.
"""Unit tests for image noise addition."""
from __future__ import annotations
import numpy as np
import pytest
import sigima.objects
import sigima.proc.image
from sigima.objects import (
NormalDistribution2DParam,
PoissonDistribution2DParam,
UniformDistribution2DParam,
)
from sigima.tests import guiutils
from sigima.tests.helpers import check_array_result, check_scalar_result
@pytest.mark.validation
def test_image_add_gaussian_noise() -> None:
"""Test :py:func:`sigima.proc.image.add_gaussian_noise`."""
# Generate a clean image.
size = 512
param = sigima.objects.Gauss2DParam.create(height=size, width=size)
ima = sigima.objects.create_image_from_param(param)
# Add Gaussian noise.
# Run twice with same parameters to check reproducibility.
p = NormalDistribution2DParam.create(seed=42, mu=0.0, sigma=1.0)
res1 = sigima.proc.image.add_gaussian_noise(ima, p)
res2 = sigima.proc.image.add_gaussian_noise(ima, p)
guiutils.view_images_side_by_side_if_gui(
[ima, res1, res2],
["Clean", "Noisy (1)", "Noisy (2)"],
title=f"Gaussian Noise Addition ({size}x{size}): "
f"Noisy images should be identical (same seed)",
)
# Shape must be preserved.
assert ima.data is not None
assert res1.data is not None
assert res1.data.shape == ima.data.shape
# Check noise statistics.
noise = res1.data - ima.data
mean_noise = float(np.mean(noise))
assert p.mu is not None
assert p.sigma is not None
expected_error = 5.0 * p.sigma / np.sqrt(ima.data.size)
check_scalar_result("Mean noise", mean_noise, p.mu, atol=expected_error)
# Identical results for same seed and distribution parameters.
assert res2.data is not None
check_array_result("Reproducibility", res1.data, res2.data)
@pytest.mark.validation
def test_image_add_poisson_noise() -> None:
"""Test :py:func:`sigima.proc.image.add_poisson_noise`."""
# Generate a clean image.
size = 512
param = sigima.objects.Gauss2DParam.create(height=size, width=size)
ima = sigima.objects.create_image_from_param(param)
# Add Poisson noise.
# Run twice with same parameters to check reproducibility.
p = PoissonDistribution2DParam.create(seed=42, lam=10.0)
res1 = sigima.proc.image.add_poisson_noise(ima, p)
res2 = sigima.proc.image.add_poisson_noise(ima, p)
guiutils.view_images_side_by_side_if_gui(
[ima, res1, res2],
["Clean", "Noisy (1)", "Noisy (2)"],
title=f"Poisson Noise Addition ({size}x{size}): "
f"Noisy images should be identical (same seed)",
)
# Shape must be preserved.
assert ima.data is not None
assert res1.data is not None
assert res1.data.shape == ima.data.shape
# Check noise statistics.
noise = res1.data - ima.data
mean_noise = float(np.mean(noise))
assert p.lam is not None
# For Poisson distribution, variance equals the mean (λ)
expected_error = 5.0 * np.sqrt(p.lam) / np.sqrt(ima.data.size)
check_scalar_result("Mean noise", mean_noise, p.lam, atol=expected_error)
# Identical results for same seed and distribution parameters.
assert res2.data is not None
check_array_result("Reproducibility", res1.data, res2.data)
@pytest.mark.validation
def test_image_add_uniform_noise() -> None:
"""Test :py:func:`sigima.proc.image.add_uniform_noise`."""
# Generate a clean image.
size = 512
param = sigima.objects.Gauss2DParam.create(height=size, width=size)
ima = sigima.objects.create_image_from_param(param)
# Add uniform noise.
# Run twice with same parameters to check reproducibility.
p = UniformDistribution2DParam.create(seed=42, vmin=-1.0, vmax=1.0)
res1 = sigima.proc.image.add_uniform_noise(ima, p)
res2 = sigima.proc.image.add_uniform_noise(ima, p)
guiutils.view_images_side_by_side_if_gui(
[ima, res1, res2],
["Clean", "Noisy (1)", "Noisy (2)"],
title=f"Uniform Noise Addition ({size}x{size}): "
f"Noisy images should be identical (same seed)",
)
# Shape must be preserved.
assert ima.data is not None
assert res1.data is not None
assert res1.data.shape == ima.data.shape
# Check noise statistics.
noise = res1.data - ima.data
mean_noise = float(np.mean(noise))
assert p.vmin is not None
assert p.vmax is not None
expected_mean = (p.vmin + p.vmax) / 2.0
# For uniform distribution, variance = (b-a)^2 / 12
variance = ((p.vmax - p.vmin) ** 2) / 12.0
expected_error = 5.0 * np.sqrt(variance) / np.sqrt(ima.data.size)
check_scalar_result("Mean noise", mean_noise, expected_mean, atol=expected_error)
# Identical results for same seed and distribution parameters.
assert res2.data is not None
check_array_result("Reproducibility", res1.data, res2.data)
if __name__ == "__main__":
guiutils.enable_gui()
test_image_add_gaussian_noise()
test_image_add_poisson_noise()
test_image_add_uniform_noise()
|
