-
Notifications
You must be signed in to change notification settings - Fork 0
/
find_circles.py
executable file
·319 lines (248 loc) · 10.6 KB
/
find_circles.py
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
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
#!/usr/bin/python
import operator
import getopt
import random
import sys
import cv
import cv2
import math
import util
OUTER_SEARCH_RADIUS = 20
INNER_SEARCH_RADIUS = 5
MIN_CONTOUR_AREA = 20
__all__ = ['find_concentric_circles', 'find_circles']
def dist_sqr(c1, c2):
return (c1[0] - c2[0]) ** 2 + (c1[1] - c2[1]) ** 2
class SpacialHash():
def __init__(self, box_size=20):
self.box_size = box_size
self.lookup = {}
def _round_coords(self, coords):
return tuple(map(lambda x: int(x) - int(x) % self.box_size, coords))
def add(self, obj, coords):
r_coords = self._round_coords(coords)
self.lookup.setdefault(r_coords, {})[obj] = coords
def delete(self, obj, coords):
r_coords = _round_coords(coords)
assert obj in self.lookup[r_coords]
assert self.lookup[r_coords][obj] == coords
def search(self, coords, radius):
mins = self._round_coords((coords[0] - radius, coords[1] - radius))
maxs = self._round_coords((coords[0] + radius + self.box_size,
coords[1] + radius + self.box_size))
for y in xrange(mins[1], maxs[1], self.box_size):
for x in xrange(mins[0], maxs[0], self.box_size):
if (x, y) in self.lookup:
for obj, coords2 in self.lookup[(x, y)].iteritems():
if dist_sqr(coords, coords2) < radius**2:
yield (obj, coords2)
def __iter__(self):
for box in self.lookup.values():
for obj, coords in box.iteritems():
yield obj, coords
def clusters(self, radius, min_cluster_size=1, max_cluster_size=None):
out = {}
already_clustered = set()
for obj, coords in self:
cluster = []
for neighbour, neighbour_coords in self.search(coords, radius):
if neighbour not in already_clustered:
cluster.append((neighbour, neighbour_coords))
if max_cluster_size != None and len(cluster) > max_cluster_size:
break
already_clustered.add(neighbour)
if len(cluster) >= min_cluster_size and \
(max_cluster_size == None or len(cluster) <= max_cluster_size):
yield cluster
def random_color():
return cv.CV_RGB(random.uniform(0,255),
random.uniform(0,255),
random.uniform(0,255))
class Feature(object):
def draw(self, image):
raise NotImplementedError()
def get_centre(self):
raise NotImplementedError()
def __iter__(self):
return iter(self.get_centre())
class Ellipse(Feature):
def __init__(self, moments, origin):
self.origin = origin
# Model the ellipse on the component with the largest area
moments = list(moments)
m = moments[util.argmax(m.m00 for m in moments)]
self.centre = (m.m10 / m.m00, m.m01 / m.m00)
self.angle = 0.5 * math.atan2(2. * m.mu11, m.mu20 - m.mu02)
def variance_at_angle(angle):
s, c = math.sin(2.0 * angle), math.cos(2.0 * angle)
nu11, nu02, nu20 = (x / m.m00 for x in (m.mu11, m.mu02, m.mu20))
return 0.5 * (nu20 + nu02 + c * (nu20 - nu02) + 2. * s * nu11)
# Find the axis lengths up to a constant of proportionality.
self.axes = tuple(variance_at_angle(x)**0.5 for x in (self.angle, self.angle + 0.5 * math.pi))
# Scale the axes so that the area of the resulting ellipse matches the measured area.
scale_factor = math.sqrt(m.m00 / (math.pi * self.axes[0] * self.axes[1]))
self.axes = tuple(scale_factor * x for x in self.axes)
def draw(self, image):
coords = tuple(map(int, self.centre))
axes = tuple(map(int, self.axes))
cv.Ellipse(image,
coords,
axes,
self.angle * 180. / math.pi,
0,
360.0,
cv.CV_RGB(0, 255, 0))
def get_centre(self, image_space=False):
if image_space:
return self.centre
else:
return (self.centre[0] - self.origin[0], self.origin[1] - self.centre[1])
def get_angle(self):
return self.angle
def get_axes(self):
return self.axes
class Point(Feature):
def __init__(self, moments):
def moment_to_point(m):
return (m.m10 / m.m00, m.m01 / m.m00)
self.point = tuple(sum(x)/len(x) for x in zip(*[moment_to_point(m) for m in moments]))
def draw(self, image):
coords = tuple(map(int, self.point))
cv.Circle(image, coords, 5, cv.CV_RGB(0, 255, 0))
cv.Circle(image, coords, 7, cv.CV_RGB(255, 255, 255))
def get_centre(self):
return self.point
def find_concentric_circles(image_in, find_ellipses=False):
"""
Find concentric circles in an image. The concentric circles it finds are
solid white circles surrounded by 3 rings: A black ring, a white ring, and
a black ring.
image_in: A binary image to search for concentric circles.
Generates a pairs, each pair being the x,y coordinates of a concentric
circle.
"""
image = cv.CloneImage(image_in)
contours = cv.FindContours(image, cv.CreateMemStorage(0))
spacialHash = SpacialHash()
contourNum = 1
featureFactory = Ellipse if find_ellipses else Point
while contours:
color = random_color()
moments = cv.Moments(contours)
if moments.m00 > MIN_CONTOUR_AREA:
centre = (moments.m10 / moments.m00, moments.m01 / moments.m00)
spacialHash.add((contourNum, moments), centre)
contourNum += 1
contours = contours.h_next()
for cluster in spacialHash.clusters(OUTER_SEARCH_RADIUS,
min_cluster_size=4,
max_cluster_size=4):
if len(cluster) == 4:
c1 = cluster[0][1]
if all(dist_sqr(c1, c2) < INNER_SEARCH_RADIUS**2 for obj, c2 in cluster[1:]):
yield featureFactory((m for (o, m), c in cluster),
origin=(0.5 * image_in.width, 0.5 * image_in.height))
def read_circular_barcode(image,
ellipse,
annotate_image=None,
num_bits=6):
num_bars = num_bits * 2 + 6
samples_per_bar=10
num_samples = samples_per_bar * num_bars
centre = ellipse.get_centre(image_space=True)
axes = ellipse.get_axes()
angle = ellipse.get_angle()
def gen_sample_points():
for theta in [float(i) * 2. * math.pi / num_samples for i in xrange(num_samples)]:
x = math.cos(theta) * axes[0] * 5.5 / 4
y = math.sin(theta) * axes[1] * 5.5 / 4
x2 = x * math.cos(angle) + -y * math.sin(angle)
y2 = x * math.sin(angle) + y * math.cos(angle)
x, y = x2, y2
x = x + centre[0]
y = y + centre[1]
if annotate_image:
cv.Circle(annotate_image, (int(x), int(y)), 2, cv.CV_RGB(0, 255, 255))
yield x, y
def find_pattern(samples, pattern, offsets=None):
if offsets is None:
offsets = xrange(num_samples)
offsets = [offset % len(samples) for offset in offsets]
pattern = reduce(operator.add, ([x] * samples_per_bar for x in pattern))
max_rating = 0
max_offset = 0
for offset in offsets:
rating = 0
for pat_idx in xrange(len(pattern)):
if samples[(offset + pat_idx) % len(samples)] == pattern[pat_idx]:
rating += 1
if rating > max_rating:
max_rating = rating
max_offset = offset
return max_offset, float(max_rating) / len(pattern)
try:
samples = [image[y, x] < 128.0 for x, y in gen_sample_points()]
except cv2.error as e:
return None
offset, rating = find_pattern(samples, [False, True, True, True, True, False])
if rating < 0.9:
return None
offset = offset + samples_per_bar * 6
code = []
for bit_idx in xrange(num_bits):
zero_offset, zero_rating = find_pattern(samples, [False, True], xrange(offset - 3, offset + 3))
one_offset, one_rating = find_pattern(samples, [True, False], xrange(offset - 3, offset + 3))
if zero_rating > one_rating:
code += [False]
offset = zero_offset + 2 * samples_per_bar
else:
code += [True]
offset = one_offset + 2 * samples_per_bar
number = sum(1 << idx for idx, val in enumerate(reversed(code)) if val)
if annotate_image:
font = cv.InitFont(cv.CV_FONT_HERSHEY_SIMPLEX, 1, 1, 0, 3, 8)
cv.PutText(annotate_image, str(number), tuple(map(int, centre)), font, cv.CV_RGB(255, 0, 255))
return number, centre
def find_labelled_circles(image_in, thresh_file_name=None, annotate_image=None, find_ellipses=False):
"""
Find concentric circles in an image, which are identified with a barcode.
image_in: Image to search.
thresh_file_name: (Optional.) File to dump threshold image in.
annotate_image: (Optional.) Image to annotate with intermediate circles and
output.
"""
image = cv.CloneImage(image_in)
cv.AdaptiveThreshold(image, image, 255.,
cv.CV_ADAPTIVE_THRESH_MEAN_C,
cv.CV_THRESH_BINARY,
blockSize=31)
if thresh_file_name:
cv.SaveImage(thresh_file_name, image)
features = list(find_concentric_circles(image, find_ellipses=find_ellipses))
if annotate_image:
for i, feature in enumerate(features):
feature.draw(annotate_image)
out = {}
for f in features:
b = read_circular_barcode(image, f, annotate_image)
if b != None:
out[b[0]] = f
return out
if __name__ == "__main__":
optlist, args = getopt.getopt(sys.argv[1:], 'i:o:t:')
in_file_name = None
out_file_name = None
thresh_file_name = None
for opt, param in optlist:
if opt == "-i":
in_file_name = param
if opt == "-o":
out_file_name = param
if opt == "-t":
thresh_file_name = param
if not in_file_name or not out_file_name:
raise Exception("Usage: %s -i <input image> -o <output image> [-t <output threshold image>]" % sys.argv[0])
image = cv.LoadImage(in_file_name, False)
color_image = cv.LoadImage(in_file_name, True)
print find_labelled_circles(image, thresh_file_name, color_image, find_ellipses=True)
cv.SaveImage(out_file_name, color_image)