forked from twtygqyy/pytorch-vdsr
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path自動數據集製作.py
238 lines (166 loc) · 7 KB
/
自動數據集製作.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
# -*- coding: utf-8 -*-
"""
Created on Sat Jan 23 20:16:56 2021
@author: user
"""
"""
自動數據集製作
包含 插值
轉y值
讀圖
寫入h5檔案
"""
import numpy as np
import cv2
import h5py #HDF5的读取：
import time
def image_segmentation(file,pix):
global first,data_count,label_count,img_file,h5_file
big_img = cv2.imread(file)
#cv2.imshow('big Image', big_img)
big_img=cv2.resize(big_img, (800, 800), interpolation=cv2.INTER_CUBIC) #妥協方法 都縮到1000 避免存h5檔案報錯
# cv2.imshow('Image_gt', big_img)#---------------------------------show
#1000 40
# h=big_img.shape[0]
# w=big_img.shape[1]
# 裁切圖片
print("+++++++++++"+file+"+++++++++++")
for x in range(1,700,200):
for y in range(1,700,200):
img = big_img[y:y+41, x:x+41]
#img=cv2.resize(img, (1000, 1000), interpolation=cv2.INTER_CUBIC) #妥協方法 都縮到1000 避免存h5檔案報錯
h=img.shape[0]
w=img.shape[1]
img_gt_y = cv2.cvtColor(img, cv2.COLOR_BGR2YCR_CB) #轉 YCR_CB
img_gt_y = img_gt_y[:,:,0].astype(float) #得到y通道
# print(img_gt_y)
img_gt_y=img_gt_y/255 # /255在儲存 原圖
# print(img_gt_y)
# cv2.imshow('My Image_gt', img_gt_y)#---------------------------------show
# print("=======",img_gt_y.shape)
img_gt_y=img_gt_y[np.newaxis,np.newaxis,:,:] #增加維度
# print("=============",img_gt_y.shape)
img = cv2.resize(img, (int(w/2), int(h/2)), interpolation=cv2.INTER_CUBIC) #雙三次降採樣
img = cv2.resize(img, (w,h), interpolation=cv2.INTER_CUBIC) #resize(img, 寬, 高) #雙三次升採樣
#cv2.imshow('My Image_bic', img)
# time.sleep(2000)
img_y = cv2.cvtColor(img, cv2.COLOR_BGR2YCR_CB) #轉 YCR_CB
img_y = img_y[:,:,0].astype(float) #得到y通道
img_x4_y=img_y/255 # /255在儲存
# print(img_x4_y)
# cv2.imshow('My Image_bic', img_x4_y)#---------------------------------show
# print("=======",img_x4_y.shape)
img_x4_y=img_x4_y[np.newaxis,np.newaxis,:,:] #增加維度
# print("=============",img_x4_y.shape)
# return (img_gt_y,img_x4_y) #回傳圖 原圖_y值圖 , 降採樣x4_y值圖
# cv2.waitKey(0)
# cv2.destroyAllWindows()
# time.sleep(2000)
if first==True:
first=False
create_h5_file(h5_file,img_x4_y,img_gt_y)
data_count+=1
label_count+=1
continue
else:
extend_dataset(h5_file,1,'data')
extend_dataset(h5_file,1,'label')
add_image_data(h5_file,img_x4_y,img_gt_y)
data_count+=1
label_count+=1
def y_img_make(file): #讀取圖片並轉呈訓練資料
img = cv2.imread(file)
#print(img.shape[0],img.shape[1],img.shape[2])#高 寬 通道
img=cv2.resize(img, (1000, 1000), interpolation=cv2.INTER_CUBIC) #妥協方法 都縮到1000 避免存h5檔案報錯
h=img.shape[0]
w=img.shape[1]
img_gt_y = cv2.cvtColor(img, cv2.COLOR_BGR2YCR_CB) #轉 YCR_CB
img_gt_y = img[:,:,0].astype(float) #得到y通道
img_gt_y=img_gt_y/255 # /255在儲存 原圖
print("=======",img_gt_y.shape)
img_gt_y=img_gt_y[np.newaxis,np.newaxis,:,:] #增加維度
print("=============",img_gt_y.shape)
img = cv2.resize(img, (int(w/4), int(h/4)), interpolation=cv2.INTER_CUBIC) #雙三次降採樣
img = cv2.resize(img, (w,h), interpolation=cv2.INTER_CUBIC) #resize(img, 寬, 高) #雙三次升採樣
img_y = cv2.cvtColor(img, cv2.COLOR_BGR2YCR_CB) #轉 YCR_CB
img_y = img_y[:,:,0].astype(float) #得到y通道
img_x4_y=img_y/255 # /255在儲存
print("=======",img_x4_y.shape)
img_x4_y=img_x4_y[np.newaxis,np.newaxis,:,:] #增加維度
print("=============",img_x4_y.shape)
return (img_gt_y,img_x4_y) #回傳圖 原圖_y值圖 , 降採樣x4_y值圖
def h5_print(file):
f = h5py.File(file,'r') #打开h5文件 # 可以查看所有的主键
for key in f.keys():
print(f[key].name)
print(f[key].shape)
#print(f[key].value)
# frame=np.zeros((2,1,41,41))
# a=np.zeros((1,4))
# b=np.zeros((1,4))
# b=((1,2,3,4))
# initial=> create h5
def create_h5_file(file,data,label):
with h5py.File(file, "w") as f:
dset = f.create_dataset('data', data=data, maxshape=(None, 1,None,None), chunks=True)
dset = f.create_dataset('label', data=label,maxshape=(None, 1,None,None), chunks=True)
# dset = f.create_dataset('test', data=a, maxshape=(None, 4), chunks=True)
# print(f['test'].value)
f.close()
# extend dataset
def extend_dataset(file,num,key): #擴展資料
with h5py.File(file, "a") as f:
f[key].resize((f[key].shape[0] + num), axis=0)
# f['test'].resize((f['test'].shape[0] + num), axis=0)
# f['test'] [2]= b
# print(f['test'].value)
# print("====",f['test'].shape[0] )
# hf[key][-1:] = frame
f.close()
def label_conut(file,key): #暫時不寫
with h5py.File(file, "a") as f:
print(key,"=" , f[key].shape[0])
def add_image_data(file,data,label): #放入1筆資料
global data_count
global label_count
print("data_count=",data_count,"label_count=",label_count)
with h5py.File(file, "a") as f:
f['data'] [data_count]= data
f['label'] [label_count]= label
f.close()
#=====================================
img_file="C:/Users/user/Desktop/Flickr2K_vdsr_train_data/Flickr2K/Flickr2K_HR/"
h5_file='D:/mytestfile_41x41_all_small_x2.h5' #mytestfile.hdf5
data_count=0
label_count=0
first=True
segment=True #s需要分割
# h5_print('D:/mytestfile.h5')
if __name__ == "__main__":
for i in range(1,2649):
print("===第",i,"張圖片")
if (i//10 ==0 ):
img_file_no="00000"+str(i)+".png"
elif (i//100 ==0 ):
img_file_no="0000"+str(i)+".png"
elif (i//1000 ==0 ):
img_file_no="000"+str(i)+".png"
elif (i//10000==0 ):
img_file_no="00"+str(i)+".png"
print(img_file_no+"\n")
if segment==False:
img_gt_y,img_x4_y=y_img_make(img_file+img_file_no)
if first==True:
first=False
create_h5_file(h5_file,img_x4_y,img_gt_y)
data_count+=1
label_count+=1
continue
else:
extend_dataset(h5_file,1,'data')
extend_dataset(h5_file,1,'label')
add_image_data(h5_file,img_x4_y,img_gt_y)
data_count+=1
label_count+=1
elif segment==True:
image_segmentation(img_file+img_file_no,40)