Merge branch 'replicating_domijan_stimuli' into 'master'

added tests to compare with hardcoded domijan stimuli

See merge request computational-psychology/stimuli!8

tests/papers/domijan2015/ground_truth.py

@@ -0,0 +1,243 @@
+import numpy as np
+
+lum_white = 9.0
+lum_black = 1.0
+lum_gray = 5.0
+
+def get_dungeon():
+    # Dungeon illusion (Bressan, 2001)
+    img = lum_black * np.ones([110, 220])
+    img[:, 110:220] = lum_white
+
+    for i in range(9, 90, 20):
+        for j in range(9, 90, 20):
+            img[i : i + 10, j : j + 10] = lum_white
+        for j in range(119, 210, 20):
+            img[i : i + 10, j : j + 10] = lum_black
+
+    img[49:59, 29:39] = lum_gray
+    img[49:59, 49:59] = lum_gray
+    img[49:59, 69:79] = lum_gray
+    img[49:59, 139:149] = lum_gray
+    img[49:59, 159:169] = lum_gray
+    img[49:59, 179:189] = lum_gray
+
+    img[29:39, 49:59] = lum_gray
+    img[69:79, 49:59] = lum_gray
+    img[29:39, 159:169] = lum_gray
+    img[69:79, 159:169] = lum_gray
+
+    return img
+
+def get_cube():
+    # Cube illusion (Agostini & Galmonte, 2002)
+    img = lum_black * np.ones([100, 200])
+
+    img[:, 100:200] = lum_white
+    img[9:20, 9:90] = lum_white
+    img[9:90, 9:20] = lum_white
+    img[79:90, 9:90] = lum_white
+    img[9:90, 79:90] = lum_white
+
+    img[9:20, 109:190] = lum_black
+    img[9:90, 109:120] = lum_black
+    img[79:90, 109:190] = lum_black
+    img[9:90, 179:190] = lum_black
+
+    img[27:32, 9:90] = lum_black
+    img[47:52, 9:90] = lum_black
+    img[67:72, 9:90] = lum_black
+
+    img[9:90, 27:32] = lum_black
+    img[9:90, 47:52] = lum_black
+    img[9:90, 67:72] = lum_black
+
+    img[32:47, 9:20] = lum_gray
+    img[52:67, 79:90] = lum_gray
+    img[79:90, 32:47] = lum_gray
+    img[9:20, 52:67] = lum_gray
+
+    img[27:32, 109:190] = lum_white
+    img[47:52, 109:190] = lum_white
+    img[67:72, 109:190] = lum_white
+
+    img[9:90, 127:132] = lum_white
+    img[9:90, 147:152] = lum_white
+    img[9:90, 167:172] = lum_white
+
+    img[32:47, 109:120] = lum_gray
+    img[52:67, 179:190] = lum_gray
+    img[79:90, 132:147] = lum_gray
+    img[9:20, 152:167] = lum_gray
+
+    return img
+
+def get_grating():
+    # Grating illusion
+    img = lum_black * np.ones([100, 220])
+    img[:, 110:220] = lum_white
+
+    for j in range(9, 100, 20):
+        img[9:90, j : j + 10] = lum_white
+
+    for j in range(119, 210, 20):
+        img[9:90, j : j + 10] = lum_black
+
+    img[9:90, 49:59] = lum_gray
+    img[9:90, 159:169] = lum_gray
+
+    return img
+
+def get_ring():
+    # Ring pattern
+    img = lum_black * np.ones([100, 200])
+    img[9:90, 9:90] = lum_white
+    img[14:85, 14:85] = lum_black
+    img[19:80, 19:80] = lum_white
+    img[24:75, 24:75] = lum_gray
+    img[29:70, 29:70] = lum_white
+    img[34:65, 34:65] = lum_black
+    img[39:60, 39:60] = lum_white
+    img[44:55, 44:55] = lum_black
+
+    img[9:90, 109:190] = lum_white
+    img[14:85, 114:185] = lum_black
+    img[19:80, 119:180] = lum_white
+    img[24:75, 124:175] = lum_black
+    img[29:70, 129:170] = lum_gray
+    img[34:65, 134:165] = lum_black
+    img[39:60, 139:160] = lum_white
+    img[44:55, 144:155] = lum_black
+
+    return img
+
+def get_bullseye():
+    # Bullseye illusion
+    img = lum_black * np.ones([100, 200])
+    img[9:90, 9:90] = lum_white
+    img[14:85, 14:85] = lum_black
+    img[19:80, 19:80] = lum_white
+    img[24:75, 24:75] = lum_black
+    img[29:70, 29:70] = lum_white
+    img[34:65, 34:65] = lum_black
+    img[39:60, 39:60] = lum_white
+    img[44:55, 44:55] = lum_gray
+
+    img[14:85, 114:185] = lum_white
+    img[19:80, 119:180] = lum_black
+    img[24:75, 124:175] = lum_white
+    img[29:70, 129:170] = lum_black
+    img[34:65, 134:165] = lum_white
+    img[39:60, 139:160] = lum_black
+    img[44:55, 144:155] = lum_gray
+
+    return img
+
+def get_sbc():
+    # Simultaneous brightness contrast
+    img = lum_black * np.ones([100, 200])
+    img[:, 0:100] = lum_white
+    img[39:60, 39:60] = lum_gray
+    img[39:60, 139:160] = lum_gray
+
+    return img
+
+def get_white():
+    # White illusion
+    img = lum_gray * np.ones([100, 100])
+    img[9:90, 9:19] = lum_black
+    img[9:90, 19:29] = lum_white
+    img[9:90, 29:39] = lum_black
+    img[9:90, 39:49] = lum_white
+    img[9:90, 49:59] = lum_black
+    img[9:90, 59:69] = lum_white
+    img[9:90, 69:79] = lum_black
+    img[9:90, 79:89] = lum_white
+    img[39:60, 29:39] = lum_gray
+    img[39:60, 59:69] = lum_gray
+
+    return img
+
+def get_benary():
+    # Benary's cross
+    img = lum_white * np.ones([100, 100])
+    img[39:60, 9:90] = lum_black
+    img[9:90, 39:60] = lum_black
+    img[39:50, 79:90] = lum_gray
+    img[28:39, 28:39] = lum_gray
+
+    return img
+
+def get_todorovic():
+    # Todorovic's illusion
+    img = lum_white * np.ones([100, 200])
+    img[:, 0:100] = lum_black
+    img[29:70, 29:70] = lum_gray
+    img[29:70, 129:170] = lum_gray
+
+    img[14:45, 14:45] = lum_white
+    img[14:45, 54:85] = lum_white
+    img[54:85, 14:45] = lum_white
+    img[54:85, 54:85] = lum_white
+
+    img[14:45, 114:145] = lum_black
+    img[14:45, 154:185] = lum_black
+    img[54:85, 114:145] = lum_black
+    img[54:85, 154:185] = lum_black
+
+    return img
+
+def get_contrast_contrast():
+    # Contrast-contrast effect
+    img = lum_gray * np.ones([100, 200])
+    img[9:89, 9:89] = lum_black
+
+    for i in range(9, 80, 20):
+        for j in range(19, 80, 20):
+            img[i : i + 10, j : j + 10] = lum_white
+            img[j : j + 10, i : i + 10] = lum_white
+
+    img[29:69, 29:69] = (lum_white + lum_gray) / 2.0
+    for i in range(29, 60, 20):
+        for j in range(29, 60, 20):
+            img[i : i + 10, j : j + 10] = (lum_black + lum_gray) / 2.0
+            k, l = i + 10, j + 10
+            img[k : k + 10, l : l + 10] = (lum_black + lum_gray) / 2.0
+
+    img[29:69, 129:169] = img[29:69, 29:69]
+
+    return img
+
+def get_checkerboard():
+    # Checkerboard contrast
+    img = lum_gray * np.ones([100, 100])
+    img[9:89, 9:89] = lum_white
+
+    for i in range(9, 80, 20):
+        for j in range(9, 80, 20):
+            img[i : i + 10, j : j + 10] = lum_black
+            k, l = i + 10, j + 10
+            img[k : k + 10, l : l + 10] = lum_black
+
+    img[39:49, 29:39] = lum_gray
+    img[59:69, 59:69] = lum_gray
+
+    return img
+
+def get_checkerboard_extended():
+    # Extended version of checkerboard contrast
+    img = lum_gray * np.ones([100, 100])
+    img[9:89, 9:89] = lum_white
+
+    for i in range(9, 80, 20):
+        for j in range(9, 80, 20):
+            img[i : i + 10, j : j + 10] = lum_black
+            k, l = i + 10, j + 10
+            img[k : k + 10, l : l + 10] = lum_black
+
+    img[39:49, 19:49] = lum_gray
+    img[59:69, 49:79] = lum_gray
+    img[29:59, 29:39] = lum_gray
+    img[49:79, 59:69] = lum_gray
+
+    return img

tests/papers/domijan2015/test_stimuli.py

@@ -0,0 +1,64 @@
+import pytest
+import numpy as np
+import matplotlib.pyplot as plt
+
+import ground_truth
+import stimuli.papers.domijan2015
+from stimuli.utils import compare_plots
+
+
+def test_dungeon():
+    original = ground_truth.get_dungeon()
+    test = stimuli.papers.domijan2015.dungeon().img
+    assert np.array_equal(original, test)
+
+
+def test_cube():
+    original = ground_truth.get_cube()
+    test = stimuli.papers.domijan2015.cube().img
+    assert np.array_equal(original, test)
+
+def test_grating():
+    original = ground_truth.get_grating()
+    test = stimuli.papers.domijan2015.grating().img
+    assert np.array_equal(original, test)
+
+def test_ring():
+    original = ground_truth.get_ring()
+    test = stimuli.papers.domijan2015.rings().img
+    assert np.array_equal(original, test)
+
+def test_bullseye():
+    original = ground_truth.get_bullseye()
+    test = stimuli.papers.domijan2015.bullseye().img
+    assert np.array_equal(original, test)
+
+def test_sbc():
+    original = ground_truth.get_sbc()
+    test = stimuli.papers.domijan2015.simultaneous_brightness_contrast().img
+    assert np.array_equal(original, test)
+
+def test_white():
+    original = ground_truth.get_white()
+    test = stimuli.papers.domijan2015.white().img
+    assert np.array_equal(original, test)
+
+def test_benary():
+    original = ground_truth.get_benary()
+    test = stimuli.papers.domijan2015.benary().img
+    assert np.array_equal(original, test)
+
+def test_todorovic():
+    original = ground_truth.get_todorovic()
+    test = stimuli.papers.domijan2015.todorovic().img
+    assert np.array_equal(original, test)
+
+def test_checkerboard():
+    original = ground_truth.get_checkerboard()
+    test = stimuli.papers.domijan2015.checkerboard().img
+    assert np.array_equal(original, test)
+
+def test_checkerboard_extended():
+    original = ground_truth.get_checkerboard_extended()
+    test = stimuli.papers.domijan2015.checkerboard_extended().img
+    assert np.array_equal(original, test)