-
Notifications
You must be signed in to change notification settings - Fork 0
/
analyse.py
463 lines (380 loc) · 10.8 KB
/
analyse.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
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
import hues
from genimg import gen_hs
from logging import info
import copy
from math import sqrt
from pdb import set_trace
# analysis params
nmap = 16
maxn = 0
focus = 0.8
limit = 0.0
dmax = 0.3
# training params
emph = 1.0
mu = 0.8
niter = 500
dlimit = 0
# other parms
logstep = 50
contrast = 3.0
def gen_linmap(n):
ver = [[float(x)/n for x in range(n)] for y in range(n)]
hor = [[float(y)/n for x in range(n)] for y in range(n)]
return (hor, ver)
def _sample(hsv, hs_map):
global maxn
hn,sn,vn = hsv[0], hsv[1], hsv[2]
# indices
hi = int(round(hn*(nmap-1)))
si = int(round(sn*(nmap-1)))
hs_map[hi][si] += 1
maxn = max(maxn, hs_map[hi][si])
def create_histogram(hsv_data):
# setup map
hs_map = [[0 for i in range(nmap)] for j in range(nmap)]
# sample data
for col in hsv_data:
_sample(col, hs_map)
info('Max n: %d', maxn)
# generate map hsv values
vals = [[(float(v)/maxn)**focus for v in row] for row in hs_map]
return vals
def act(diff):
return ((diff + 1.0)/2.0)**emph
'''
Select sub-matrix from square matrix 'm' at
row/column i,j of size 3x3.
Border cases are either zero or column
wrapped if rot_j=True.
Returns tuple ([m[i-1][j-1]..m[i+1][j+1]],
m[i][j])
'''
def select(m, i,j, rot_j=True):
nr = len(m)
nc = len(m[0])
if nr != nc:
raise ValueError('m must be square')
vals = []
for row in range(i-1,i+2):
for col in range(j-1,j+2):
if rot_j: col = col % nc
if row<0 or row>=nr: val = 0.0
elif col<0 or col>=nc: val=0.0
else: val=m[row][col]
if not(row==i and col==j):
vals.append(val)
return vals, m[i][j]
'''
return histogram with extrema set to 1.0.
If cents=true, also returns centroids as
arrays of hues and saturations plus array
of centroid weights.
NOTE: fuer filter wird nur huec/satc benoetigt
'''
def extrema(hist, cents=False):
ext = copy.copy(hist)
(hues, sats) = gen_linmap(nmap)
huec, satc, wc = [], [], []
for i in range(len(hist)):
for j in range(len(hist)):
(vals, v) = select(hist, i,j)
# edge detection
sv = sorted(vals)
if v - sum(sv[:4])/4 > dmax:
wc.append(v)
huec.append(hues[i][j])
satc.append(sats[i][j])
ext[i][j] = 1.0
else:
ext[i][j] = 0.0
if cents:
return ext, huec, satc, wc
else:
return ext
'''
Return flattened list of new values from
vals matching frequency in freq.
vals must be sorted!
'''
def flatten(vals, freq):
hist = sorted(freq)
cum_dist = []
sumf = 0
for x in hist:
sumf += x
cum_dist.append(sumf)
return cum_dist
def _update_xform(hues, sats, vals):
sumd = 0.0
new_hues = copy.copy(hues)
new_sats = copy.copy(sats)
for i in range(nmap):
for j in range(nmap):
# hues
hue = hues[i][j]
val = vals[i][j]
jm1 = (j - 1) % nmap # Spalte rotiert bei hue
jp1 = (j + 1) % nmap
im1 = max(i - 1, 0)
ip1 = min(i + 1, nmap-1)
#dhue = (hues[im1][j]-hue)*vals[im1][j] + (hues[i][jm1]-hue)*vals[i][jm1] + (hues[ip1][j]-hue)*vals[ip1][j] + (hues[i][jp1]-hue)*vals[i][jp1]
dhue = (hues[im1][j]-hue)*vals[im1][j]**emph + (hues[i][jm1]-hue)*vals[i][jm1]**emph + (hues[ip1][j]-hue)*vals[ip1][j]**emph + (hues[i][jp1]-hue)*vals[i][jp1]**emph
new_hues[i][j] += (1-vals[i][j]) * dhue * mu
sumd += dhue * mu * (1-vals[i][j])
# sats
jm1 = max(j - 1, 0) # Bei saturation rotiert Spalte nicht!
jp1 = min(j + 1, nmap-1)
sat = sats[i][j]
#dsat = (sats[im1][j]-sat)*vals[im1][j] + (sats[i][jm1]-sat)*vals[i][jm1] + (sats[ip1][j]-sat)*vals[ip1][j] + (sats[i][jp1]-sat)*vals[i][jp1]
dsat = (sats[im1][j]-sat)*vals[im1][j]**emph + (sats[i][jm1]-sat)*vals[i][jm1]**emph + (sats[ip1][j]-sat)*vals[ip1][j]**emph + (sats[i][jp1]-sat)*vals[i][jp1]**emph
new_sats[i][j] += dsat * mu * (1-vals[i][j])
return new_hues, new_sats, sumd
def _update_xform_max(hues, sats, vals):
sumd = 0.0
new_hues = copy.copy(hues)
new_sats = copy.copy(sats)
for i in range(nmap):
for j in range(nmap):
(surr_vals, val) = select(vals, i,j)
if val==1.0: continue
max_v = max(surr_vals)
idx_v = surr_vals.index(max_v)
(surr_hues, hue) = select(hues, i,j)
(surr_sats, sat) = select(sats, i,j, rot_j=False)
mhue = surr_hues[idx_v]
msat = surr_sats[idx_v]
dhue = (mhue-hue)*max_v
dsat = (msat-sat)*max_v
new_hues[i][j] += (1-val) * dhue * mu
new_sats[i][j] += (1-val) * dsat * mu
sumd += dhue * mu * (1-val)
return new_hues, new_sats, sumd
def rdist(x,h):
d = x - h
if abs(d) > 0.5:
d = d - d/abs(d)
return d
def memb(hue,sat, huec,satc):
hdist = [rdist(hue,c) for c in huec]
sdist = [sat-c for c in satc]
dist = [rdist(hue,c)**2+(sat-s)**2 for (c,s) in zip(huec,satc)]
u = [1/sum([((di+0.0001)/(dj+0.0001))**(2.0/(contrast-1.0)) for dj in dist]) for di in dist]
return u, hdist, sdist
def _xform(hsv, hues, sats, hilite=False):
hn,sn,vn = hsv[0], hsv[1], hsv[2]
# indices
hi = int(round(hn*(nmap-1)))
si = int(round(sn*(nmap-1)))
hue = hues[hi][si]
sat = sats[hi][si]
if hilite:
d = ((hue-hn)**2 + (sat-sn)**2)**0.5 / 2**0.5
vn = d*nmap/4
return (hue, sat, vn)
def _xform_vals(hsv, vals):
hn,sn,vn = hsv[0], hsv[1], hsv[2]
# indices
hi = int(round(hn*(nmap-1)))
si = int(round(sn*(nmap-1)))
fq = vals[hi][si]
if fq>limit:
pass
#sn *= fq
#vn *= fq**2
else:
sn = 0.0
vn = 0.0
return (hn, sn, vn)
def train_map(vals, updt_fct=_update_xform):
(hues, sats) = gen_linmap(nmap)
info('Start training')
for i in range(niter):
hues, sats, sumd = updt_fct(hues, sats, vals)
if (i+1) % logstep == 0:
info('Update iteration %d, %f', i+1, sumd)
if abs(sumd) < float(dlimit)/nmap: break
return hues, sats
def adapt(hsv, hues, sats, xfct=_xform, hilite=False):
hsv_new = [xfct(c, hues, sats, hilite) for c in hsv]
return hsv_new
def adapt_vals(hsv, vals):
hsv_new = [_xform_vals(c, vals) for c in hsv]
return hsv_new
'''
Unit tests
'''
import unittest
test_all = False
class TestColorAnalysis(hues.PaletteTestBase):
# wie besser?
@classmethod
def setUpClass(cls):
hues.PaletteTestBase.setUpClass()
def setUp(self):
self.hsv_data = hues.PaletteTestBase.get_hsv('pond')
self.hsv_palette = hues.PaletteTestBase.get_hsv('palette')
self.hsv_test = hues.PaletteTestBase.get_hsv('kueche')
@unittest.skipUnless(test_all, 'test_all not set')
def test_histogram(self):
vals = create_histogram(self.hsv_data)
img = gen_hs(vmap=vals, nc=nmap)
img.show()
@unittest.skipUnless(test_all, 'test_all not set')
def test_extrema(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
img = gen_hs(vmap=ext, nc=nmap)
img.show()
@unittest.skipUnless(test_all, 'test_all not set')
def test_xform(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
img0 = gen_hs(vmap=ext, nc=nmap)
# calc xform matrix
(hues, sats) = train_map(ext)
# visualize xform matrix
img = gen_hs(hmap=hues, smap=sats, nc=nmap, v_def=1.0)
img0.show()
img.show()
'''
Histogrammdarstellung sollte wie Original-Palette aussehen
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_identity_xform(self):
focus = 0.4
vals = create_histogram(self.hsv_palette)
# calc xform matrix
(hues, sats) = train_map(vals)
# visualize xform matrix
img = gen_hs(hmap=hues, smap=sats, vmap=vals, nc=nmap, v_def=1.0)
img.show()
'''
Transformation mit Palette sollte Bild nicht veraendern
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_identity(self):
vals = create_histogram(self.hsv_palette)
(hues, sats) = train_map(vals)
# adapt palette image
hsv_new = adapt(self.hsv_palette, hues, sats)
self.render(hsv_new, palette=False)
'''
Palette modifiziert
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_adapt_palette(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
(hues, sats) = train_map(ext)
# adapt palette image
hsv_new = adapt(self.hsv_palette, hues, sats)
self.render(hsv_new, palette=False)
'''
Foto mit eigenem Spektrum modifiziert.
Sollte Farben hoechstens verstaerken
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_adapt_self(self):
vals = create_histogram(self.hsv_data)
(hues, sats) = train_map(vals)
# adapt palette image
hsv_new = adapt(self.hsv_data, hues, sats)
self.render(hsv_new, palette=False)
'''
Foto modifiziert
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_adapt_img(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
(hues, sats) = train_map(ext)
# show map
img = gen_hs(hmap=hues, smap=sats, nc=nmap, v_def=1.0)
img.show()
# adapt test image
hsv_new = adapt(self.hsv_test, hues, sats)
self.render(hsv_new, palette=False)
'''
Foto modifiziert mit max training algo
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_adapt_img_max(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
(hues, sats) = train_map(ext, updt_fct=_update_xform_max)
# show map
img = gen_hs(hmap=hues, smap=sats, nc=nmap, v_def=1.0)
img.show()
# adapt test image
hsv_new = adapt(self.hsv_test, hues, sats)
self.render(hsv_new, palette=False)
'''
Foto modifiziert, Aenderungen hervorheben
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_highlight_changes(self):
vals = create_histogram(self.hsv_data)
ext = extrema(vals)
(hues, sats) = train_map(ext)
# show map
img = gen_hs(hmap=hues, smap=sats, nc=nmap, v_def=1.0)
img.show()
# adapt test image
hsv_new = adapt(self.hsv_test, hues, sats, hilite=True )
self.render(hsv_new, palette=False)
'''
Foto mit Hauptfarben hervorgehoben
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_highlight_palette(self):
vals = create_histogram(self.hsv_data)
(hues, sats) = gen_linmap(nmap)
# show map
img = gen_hs(hmap=hues, smap=sats, nc=nmap, vmap=vals)
img.show()
# adapt self image
hsv_new = adapt_vals(self.hsv_data, vals)
self.render(hsv_new, palette=False)
# adapt test image
hsv_new = adapt_vals(self.hsv_test, vals)
self.render(hsv_new, palette=False)
'''
Test activation function
'''
@unittest.skipUnless(test_all, 'test_all not set')
def test_memb(self):
hue = 0.3
sat = 0.6
huec = [0.0, 0.5, 0.7]
satc = [0.1, 0.2, 0.3]
u, d = memb(hue,sat,huec,satc)
print
print u
print d
self.assertEqual(sum(u), 1.0)
'''
Foto modifiziert mit cluster membership
'''
#@unittest.skipUnless(test_all, 'test_all not set')
def test_adapt_img_memb(self):
vals = create_histogram(self.hsv_data)
(ext, huec, satc, wc) = extrema(vals, cents=True)
# show extrema
img = gen_hs(vmap=ext, nc=nmap)
img.show()
# adapt test image
hsv_new = []
for hsv in self.hsv_test:
hue = hsv[0]
sat = hsv[1]
u,hdist,sdist = memb(hue,sat,huec,satc)
dhue = sum([d*m for (d,m) in zip(hdist,u)])
dsat = sum([d*m for (d,m) in zip(sdist,u)])
hue -= dhue
sat -= dsat
hsv_new.append([hue, sat, hsv[2]])
self.render(hsv_new, palette=False)
#
if __name__ == '__main__':
unittest.main()