-
Notifications
You must be signed in to change notification settings - Fork 1
/
Example8a_hyperspectral.py
182 lines (147 loc) · 6.68 KB
/
Example8a_hyperspectral.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
import pandas as pd
import numpy as np
from PowerRcut.PowerRcut_hyperspectral import PRcut
from PowerRcut.img_to_graph.img_to_graph import *
from matplotlib import pyplot as plt
from sklearn.cluster import k_means
from scipy.sparse.csgraph import connected_components, laplacian
from scipy.sparse.linalg import eigsh
import scipy as sp
from tqdm import tqdm
import time
from sklearn.metrics import adjusted_rand_score as ARI
from sklearn.metrics import adjusted_mutual_info_score as AMI
from sklearn.metrics import confusion_matrix
from sklearn.datasets import make_blobs
from sklearn.decomposition import PCA
from scipy.io import loadmat
"""
Calculating the timing statistics for different subimages of University Pavia Dataset
-------------------------------------------------------------------------------------
"""
arr_scale = []
arr_processtime = []
arr_Rcut_time = []
arr_accRcut = []
arr_PRcut_time = []
arr_accPRcut = []
for scale in np.arange(0.7,0.99,0.1):
# Need only construct the graph once per scale. (Much faster this way)
st_time = time.clock()
img = loadmat("./Hyperspectral_data/PaviaU.mat")['paviaU']
s0,s1,s2 = img.shape
img = np.array(img[-int(scale*s0):,:int(scale*s1),:],np.int32)
img_gt = loadmat("./Hyperspectral_data/PaviaU_gt.mat")['paviaU_gt']
img_gt = img_gt[-int(scale*s0):,:int(scale*s1)]
s0,s1,s2 = img.shape
# Constructing the graph from the data using the PowerRcut.img_to_graph module.
graph, w = method1(img,s0,s1,s2,spat_distance = 1, beta = 5., eps = 1e-6)
graph = 0.5 * (graph + graph.T)
process_time = time.clock() - st_time
for _ in range(10):
# Performing spectral clustering on the data
st_time = time.clock()
L = laplacian(graph)
eigval, embedRcut = eigsh(L, 20, sigma = 1e-6 )
d0, labelsRcut, d2 = k_means(embedRcut,50)
Rcut_time = time.clock() - st_time
# Calculating the accuracy for spectral clustering
# Each class obtained on clustering is assumed to belong to the largest intersection
z = confusion_matrix(img_gt.flatten(),labelsRcut)
accRcut = 0
for i in range(z.shape[0]):
accRcut += z[np.argmax(z[:,i]),i]
accRcut = (accRcut+0.0)/(s0*s1)
# Performing PRcut on the data
st_time = time.clock()
embedPRcut, label_components = PRcut(graph, 20, discrete_Method = 'kmeans', nClust=7, threshComp=200, verbose=True)
d0, labelsPRcut, d2 = k_means(embedPRcut,50)
PRcut_time = time.clock() - st_time
# Calculating the accuracy for PRcut
# Each class obtained on clustering is assumed to belong to the largest intersection
z = confusion_matrix(img_gt.flatten(),labelsPRcut)
accPRcut = 0
for i in range(z.shape[0]):
accPRcut += z[np.argmax(z[:,i]),i]
accPRcut = (accPRcut+0.0)/(s0*s1)
arr_scale.append(scale)
arr_processtime.append(process_time)
arr_Rcut_time.append(Rcut_time)
arr_accRcut.append(accRcut)
arr_PRcut_time.append(PRcut_time)
arr_accPRcut.append(accPRcut)
print '***********************************'
print ' Scale = ', img.shape
print 'Constructing graph time = ', process_time
print 'Ratio cut time = ', Rcut_time
print ' Accuracy = ', AMI(img_gt.flatten(),labelsRcut), ARI(img_gt.flatten(),labelsRcut),accRcut
print 'PowerRatio cut time = ', PRcut_time
print ' Accuracy = ', AMI(img_gt.flatten(),labelsPRcut), ARI(img_gt.flatten(),labelsPRcut),accPRcut
print '***********************************'
# Writing the results to a csv file
f = open('results_8a.csv', "w+")
for i in range(len(arr_scale)):
l = ""
l += str(arr_scale[i])+","+str(arr_processtime[i])+","
l += str(arr_Rcut_time[i])+","+str(arr_accRcut[i])+","+str(AMI(img_gt.flatten(),labelsRcut))+","+str(ARI(img_gt.flatten(),labelsRcut))+","
l += str(arr_PRcut_time[i])+","+str(arr_accPRcut[i])+","+str(AMI(img_gt.flatten(),labelsPRcut))+","+str(ARI(img_gt.flatten(),labelsPRcut))
f.write(l)
f.write("\n")
f.close()
"""
Recording an typical solution
-----------------------------
"""
st_time = time.clock()
img = loadmat("./Hyperspectral_data/PaviaU.mat")['paviaU']
img = img[-200:,:200,:]
img = np.array(img, dtype = np.int32)
img_gt = loadmat("./Hyperspectral_data/PaviaU_gt.mat")['paviaU_gt']
img_gt = img_gt[-200:,:200]
print '... Constructing graph'
s0,s1,s2 = img.shape
graph, w = method1(img,s0,s1,s2,spat_distance = 1, beta = 5., eps = 1e-6)
graph = 0.5 * (graph + graph.T)
process_time = time.clock() - st_time
st_time = time.clock()
L = laplacian(graph, normed = True)
eigval, embedRcut = eigsh(L, k=20, sigma = 1e-6 )
d0, labelsRcut, d2 = k_means(embedRcut,50)
Rcut_time = time.clock() - st_time
z = confusion_matrix(img_gt.flatten(),labelsRcut)
accRcut = 0
for i in range(z.shape[0]):
accRcut += z[np.argmax(z[:,i]),i]
accRcut = (accRcut+0.0)/(s0*s1)
st_time = time.clock()
embedPRcut, label_components = PRcut(graph, 20, discrete_Method = 'kmeans', nClust=7, threshComp=200, verbose=True)
d0, labelsPRcut, d2 = k_means(embedPRcut,50)
PRcut_time = time.clock() - st_time
z = confusion_matrix(img_gt.flatten(),labelsPRcut)
accPRcut = 0
for i in range(z.shape[0]):
accPRcut += z[np.argmax(z[:,i]),i]
accPRcut = (accPRcut+0.0)/(s0*s1)
print '***********************************'
print ' Scale = ', img.shape
print 'Constructing graph time = ', process_time
print 'Ratio cut time = ', Rcut_time
print ' Accuracy = ', AMI(img_gt.flatten(),labelsRcut), ARI(img_gt.flatten(),labelsRcut),accRcut
print 'PowerRatio cut time = ', PRcut_time
print ' Accuracy = ', AMI(img_gt.flatten(),labelsPRcut), ARI(img_gt.flatten(),labelsPRcut),accPRcut
print '***********************************'
plt.imsave("./img/paviaU_gt.pdf",img_gt)
plt.imsave("./img/paviaU_PRcut.pdf",np.array(labelsPRcut).reshape(img_gt.shape))
plt.imsave("./img/paviaU_Rcut.pdf",np.array(labelsRcut).reshape(img_gt.shape))
from scipy.ndimage.filters import gaussian_filter
filtered_PRcut = gaussian_filter(labelsPRcut.reshape(img_gt.shape), 2)
plt.imsave("./img/paviaU_PRcutFiltered.pdf",filtered_PRcut)
from process_small_clusters import *
embedPRcut_filtered = process_small_clusters(embedPRcut,label_components,graph,200)
d0, labelsPRcut_filtered, d2 = k_means(embedPRcut_filtered,20)
z = confusion_matrix(img_gt.flatten(),labelsPRcut_filtered)
accPRcut_filtered = 0
for i in range(z.shape[0]):
accPRcut_filtered += z[np.argmax(z[:,i]),i]
accPRcut_filtered = (accPRcut_filtered+0.0)/(s0*s1)
plt.imsave("./img/paviaU_embedFilter.pdf",np.array(labelsPRcut_filtered).reshape(img_gt.shape))