-
Notifications
You must be signed in to change notification settings - Fork 3
/
find_similar_images.py
211 lines (177 loc) · 6.6 KB
/
find_similar_images.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
import cv2
import os
import math
from scipy import ndimage, misc
import numpy as np
from matplotlib import pyplot as plt
from scipy import spatial
from collections import defaultdict
def extract_features(image, vector_size=32):
#print('Hello')
try:
alg = cv2.KAZE_create()
#alg = cv2.xfeatures2d.SIFT_create()
kps = alg.detect(image)
kps = sorted(kps, key=lambda x: -x.response)[:vector_size]
kps, dsc = alg.compute(image, kps)
if(len(kps) != 0):
dsc = dsc.flatten()
needed_size = (vector_size * 64)
if dsc.size < needed_size:
dsc = np.concatenate([dsc, np.zeros(needed_size - dsc.size)])
dsc = dsc[:2048]
#print(dsc)
else:
return(np.zeros(vector_size * 64))
except cv2.error as e:
print 'Error: ', e
return None
idx = dsc.argsort()[-1:-411:-1]
binarize_dsc = np.zeros(2048)
binarize_dsc[idx[:]] = 1
return(binarize_dsc)
#return(dsc)
def allowPermute(feature_dataset,get_permute):
new_feature_dataset = np.zeros(2048)
ans = 2049
for a in range(len(get_permute)):
if(feature_dataset[get_permute[a]] != 0.0):
ans = a
break
return(ans)
def getMatch(feature_map1,feature_map2):
#intersection_cardinality = len(set.intersection(*[set(feature_map1), set(feature_map2)]))
#union_cardinality = len(set.union(*[set(feature_map1), set(feature_map2)]))
#return intersection_cardinality/float(union_cardinality)
match_value = 0
result = spatial.distance.cosine(feature_map1, feature_map2)
return(result)
#for x in range(len(feature_map1)):
# match_value = match_value + abs(feature_map1[x] - feature_map2[x])
#return(match_value)
#resizing the images
test_images = []
for root, dirnames, filenames in os.walk("./dataset/test_image"):
for filename in filenames:
filepath = os.path.join(root, filename)
test_image = ndimage.imread(filepath, mode="L")
test_image_resized = misc.imresize(test_image, (480, 480))
test_images.append(test_image_resized)
imdataset = []
for root, dirnames, filenames in os.walk("./dataset/image_dataset"):
for filename in filenames:
filepath = os.path.join(root, filename)
imdata = ndimage.imread(filepath, mode="L")
imdata_resized = misc.imresize(imdata, (300, 300))
imdataset.append(imdata_resized)
edges = []
for i in range(len(test_images)):
getImage = test_images[i]
getedges = cv2.Canny(getImage,100,200)
edges.append(getedges)
train_edges = []
for i in range(len(imdataset)):
getImage = imdataset[i]
getedges = cv2.Canny(getImage,100,200)
train_edges.append(getedges)
test_images = edges
imdataset = train_edges
trainImagePatch = []
for n in range(len(imdataset)):
getEdgeImage = imdataset[n]
gettrainImagePatch = []
for x in range(9):
for y in range(9):
image_patch = getEdgeImage[x*30:x*30+60:1, y*30:y*30+60:1]
gettrainImagePatch.append(image_patch)
trainImagePatch.append(gettrainImagePatch)
testImagePatch = []
for n in range(len(test_images)):
getEdgeImage = test_images[n]
gettestImagePatch = []
for x in range(18):
for y in range(18):
image_patch = getEdgeImage[x*24:x*24+96:1, y*24:y*24+96:1]
gettestImagePatch.append(image_patch)
testImagePatch.append(gettestImagePatch)
train_feature_dataset = []
for x in range(len(trainImagePatch)):
feature_dataset = []
for y in range(len(trainImagePatch[0])):
train_image = trainImagePatch[x][y]
getfeatures = extract_features(train_image)
feature_dataset.append(getfeatures)
train_feature_dataset.append(feature_dataset)
test_feature_dataset = []
for a in range(len(testImagePatch)):
feature_dataset = []
for b in range(len(testImagePatch[0])):
get_test_image = testImagePatch[a][b]
test_feature = extract_features(get_test_image)
feature_dataset.append(test_feature)
test_feature_dataset.append(feature_dataset)
numOfPermute = 20
kOfPermute = 3
#print(len(arrayIndex),len(arrayIndex[0]))
testarrayIndex = np.zeros([numOfPermute,len(test_feature_dataset),len(test_feature_dataset[0])])
power_array = np.zeros(3)
getMatchIndex = defaultdict(list)
power_array[0] = 1
power_array[1] = 10
power_array[2] = 100
for numpermute in range(numOfPermute):
arrayIndex = np.zeros([len(train_feature_dataset),len(train_feature_dataset[0])])
for p in range(kOfPermute):
permute_value = np.random.permutation(2048)
for x in range(len(train_feature_dataset)):
for z in range(len(train_feature_dataset[0])):
getindex = allowPermute(train_feature_dataset[x][z],permute_value)
arrayIndex[x][z] = (getindex*power_array[p]) + arrayIndex[x][z]
#print(arrayIndex[x])
for y in range(len(test_feature_dataset)):
for w in range(len(test_feature_dataset[0])):
getindex2 = allowPermute(test_feature_dataset[y][w],permute_value)
testarrayIndex[numpermute][y][w] = (getindex2*power_array[p]) + testarrayIndex[numpermute][y][w]
#print(testarrayIndex[numpermute][y])
#print([arrayIndex[0],testarrayIndex[numpermute][0]])
for g in range(len(train_feature_dataset)):
for h in range(len(train_feature_dataset[0])):
getMatchIndex[arrayIndex[g][h]].append(g)
kmin = 6
for m in range(len(test_feature_dataset)):
Match_value = np.zeros(len(imdataset))
for n in range(len(test_feature_dataset[0])):
get_test_image = testImagePatch[m][n]
for num in range(numOfPermute):
for j in range(len(getMatchIndex[testarrayIndex[num][m][n]])):
Match_value[getMatchIndex[testarrayIndex[num][m][n]][j]] = Match_value[getMatchIndex[testarrayIndex[num][m][n]][j]] + 1
idxMostmatched = Match_value.argsort()[-1:-kmin:-1]
plt.figure(m)
plt.subplot(231),plt.imshow(test_images[m],cmap = 'gray')
plt.subplot(232),plt.imshow(imdataset[idxMostmatched[0]],cmap = 'gray')
plt.subplot(233),plt.imshow(imdataset[idxMostmatched[1]],cmap = 'gray')
plt.subplot(234),plt.imshow(imdataset[idxMostmatched[2]],cmap = 'gray')
plt.subplot(235),plt.imshow(imdataset[idxMostmatched[3]],cmap = 'gray')
plt.subplot(236),plt.imshow(imdataset[idxMostmatched[4]],cmap = 'gray')
plt.show()
"""
kmin = 5
for a in range(len(test_images)):
get_test_image = test_images[a]
test_feature = extract_features(get_test_image)
Match_value = np.zeros(len(imdataset))
for z in range(len(imdataset)):
getMatchPair = feature_dataset[z]
getMatchValue = getMatch(test_feature,getMatchPair)
Match_value[z] = getMatchValue
idx = np.argsort(Match_value)
print(Match_value[idx[:kmin]])
plt.figure(a)
plt.subplot(231),plt.imshow(get_test_image,cmap = 'gray')
plt.subplot(232),plt.imshow(imdataset[idx[0]],cmap = 'gray')
plt.subplot(233),plt.imshow(imdataset[idx[1]],cmap = 'gray')
plt.subplot(234),plt.imshow(imdataset[idx[2]],cmap = 'gray')
plt.subplot(235),plt.imshow(imdataset[idx[3]],cmap = 'gray')
plt.subplot(236),plt.imshow(imdataset[idx[4]],cmap = 'gray')
plt.show()
"""