-
Notifications
You must be signed in to change notification settings - Fork 1
/
dlm.py
187 lines (157 loc) · 6.87 KB
/
dlm.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
import csv
from skimage import filters
import matplotlib.pyplot as plt
from utils.dlm_utils import csf, dlm
from joblib import dump,Parallel,delayed
from scipy.stats import gmean
import time
from scipy.ndimage import gaussian_filter
from utils.hdr_utils import hdr_yuv_read
from utils.csf_utils import csf_barten_frequency,csf_filter_block,blockwise_csf,windows_csf
import numpy as np
import glob
import pandas as pd
import os
from os.path import join
import scipy
import socket
import sys
print(float(sys.argv[1]))
if socket.gethostname().find('tacc')>0:
csv_file_vidinfo = 'fall2021_yuv_rw_info.csv'
vid_pth = '/scratch/06776/kmd1995/video/HDR_2021_fall_yuv_upscaled/fall2021_hdr_upscaled_yuv'
out_root = '/scratch/06776/kmd1995/feats/feats/hdrdlm/dlm'
else:
csv_file_vidinfo = '/home/zaixi/code/HDRproject/hdr_vmaf/python_vmaf/fall2021_yuv_rw_info.csv'
vid_pth = '/mnt/7e60dcd9-907d-428e-970c-b7acf5c8636a/fall2021_hdr_upscaled_yuv/'
out_root = '/media/zaixi/zaixi_nas/HDRproject/feats/hdrdlm/dlm'
csv_df = pd.read_csv(csv_file_vidinfo)
files = csv_df["yuv"]
ref_files = glob.glob(join(vid_pth,'4k_ref_*'))
fps = csv_df["fps"]
framenos_list = csv_df["framenos"]//3
ws =csv_df["w"]
hs = csv_df["h"]
upscaled_yuv_names = [x[:-4]+'_upscaled.yuv' for x in csv_df['yuv']]
def logit(Y):
Y = -0.99+(Y-np.amin(Y))* 1.98/(np.amax(Y)-np.amin(Y))
Y_transform = np.log((1+(Y)**3)/(1-(Y)**3))
return Y_transform
def global_exp(image,par):
if np.max(image) > 1.1:
image = image/1023
assert len(np.shape(image)) == 2
avg = np.average(image)
y = np.exp(par*(image-avg))
return y
def gen_gauss_window(lw, sigma):
sd = np.float32(sigma)
lw = int(lw)
weights = [0.0] * (2 * lw + 1)
weights[lw] = 1.0
sum = 1.0
sd *= sd
for ii in range(1, lw + 1):
tmp = np.exp(-0.5 * np.float32(ii * ii) / sd)
weights[lw + ii] = tmp
weights[lw - ii] = tmp
sum += 2.0 * tmp
for ii in range(2 * lw + 1):
weights[ii] /= sum
return weights
def local_exp(image,par,patch_size):
assert len(np.shape(image)) == 2
h, w = np.shape(image)
if np.max(image) > 1.1:
image = image/1023
avg_window = gen_gauss_window(patch_size//2, 7.0/6.0)
mu_image = np.zeros((h, w), dtype=np.float32)
image = np.array(image).astype('float32')
scipy.ndimage.correlate1d(image, avg_window, 0, mu_image, mode='constant')
scipy.ndimage.correlate1d(mu_image, avg_window, 1, mu_image, mode='constant')
y = np.exp(par*(image - mu_image))
return y
def dlm_refall_wrapper(ind,files):
ref_f = files[ind]
content = os.path.basename(ref_f).split('_')[2]
print(content)
dis_filenames = [x for x in glob.glob(join(vid_pth,"*")) if content in x]
print(dis_filenames)
Parallel(n_jobs=30,verbose=1)(delayed(dlm_video_wrapper)(ref_f,dis_f) for dis_f in dis_filenames)
def dlm_video_wrapper(ref_f,dis_f):
if(ref_f==dis_f):
print('Videos are the same')
return
basename = os.path.basename(dis_f)
dis_index = upscaled_yuv_names.index(basename)
h = 2160 #hs[dis_index]
w = 3840 #ws[dis_index]
framenos = framenos_list[dis_index]
dlm_image_wrapper(ref_f,dis_f,framenos,h,w,use_adaptive_csf=False,nonlinear = nonlinear, par = par)
return
def dlm_image_wrapper(ref_f,dis_f,framenos,h,w,use_adaptive_csf=False,nonlinear = None,par = None,adaptation='bilateral',use_non_overlapping_blocks=False,use_views=False):
ref_file_object = open(ref_f)
dis_file_object = open(dis_f)
randlist = np.arange(framenos) # np.random.randint(0,framenos,10)
score_df = pd.DataFrame([])
dis_name = os.path.splitext(os.path.basename(dis_f))[0]
output_csv = os.path.join(out_pth,dis_name+'.csv')
if(os.path.exists(output_csv) and os.path.getsize(output_csv)>100):
return
with open(output_csv,'a') as f1:
writer=csv.writer(f1, delimiter=',',lineterminator='\n',)
writer.writerow(['framenum','dlm','aim','comb'])
for framenum in range(framenos):
try:
ref_y_pq,_,_ = hdr_yuv_read(ref_file_object,framenum,h,w)
dis_y_pq,_,_ = hdr_yuv_read(dis_file_object,framenum,h,w)
except Exception as e:
f = open("dlm_yuv_reading_error.txt", "a")
f.write(dis_name+"\n")
f.close()
break
if(use_adaptive_csf==True):
# apply CSF here
if(use_non_overlapping_blocks==True): # apply CSF on non-overlapping blocks of the image
csf_filtered_ref_y_pq = blockwise_csf(ref_y_pq)
csf_filtered_dis_y_pq = blockwise_csf(dis_y_pq)
else: # sliding window; returns filtered value at center of each sliding window
csf_filtered_ref_y_pq = windows_csf(ref_y_pq,use_views=use_views,adaptation=adaptation)
csf_filtered_dis_y_pq = windows_csf(dis_y_pq,use_views=use_views,adaptation=adaptation)
# standard DLM but without CSF
dlm_val = dlm(csf_filtered_ref_y_pq,csf_filtered_dis_y_pq,use_csf=False)
elif(nonlinear == 'logit'):
logit_ref_y_pq = logit(ref_y_pq,1)
logit_dis_y_pq = logit(dis_y_pq,1)
dlm_val = dlm(logit_ref_y_pq,logit_dis_y_pq)
elif(nonlinear == 'local_exp'):
logit_ref_y_pq = local_exp(ref_y_pq,par,31)
logit_dis_y_pq = local_exp(dis_y_pq,par,31)
dlm_val1 = dlm(logit_ref_y_pq,logit_dis_y_pq)
logit_ref_y_pq = local_exp(ref_y_pq,-par,31)
logit_dis_y_pq = local_exp(dis_y_pq,-par,31)
dlm_val2 = dlm(logit_ref_y_pq,logit_dis_y_pq)
elif(nonlinear == 'global_exp'):
logit_ref_y_pq = global_exp(ref_y_pq,par)
logit_dis_y_pq = global_exp(dis_y_pq,par)
dlm_val1 = dlm(logit_ref_y_pq,logit_dis_y_pq)
logit_ref_y_pq = global_exp(ref_y_pq,-par)
logit_dis_y_pq = global_exp(dis_y_pq,-par)
dlm_val2 = dlm(logit_ref_y_pq,logit_dis_y_pq)
else:
# standard DLM
dlm_val = dlm(ref_y_pq,dis_y_pq,use_csf=True)
if dlm_val1 is not None:
row = [framenum,dlm_val1[0],dlm_val1[1],dlm_val1[2],dlm_val2[0],dlm_val2[1],dlm_val2[2]]
else:
row = [framenum,dlm_val[0],dlm_val[1],dlm_val[2]]
writer.writerow(row)
for nonlinear in ['local_exp','global_exp']:
# for par in [[0.5,1,2,5]]:
for par in [float(sys.argv[1])]:
out_pth = f'{out_root}_{nonlinear}_{par}'
if not os.path.exists(out_pth):
os.makedirs(out_pth)
Parallel(n_jobs=3,verbose=1)(delayed(dlm_refall_wrapper)(i,ref_files) for i in range(len(ref_files)))
# #for i in range(len(ref_files)):
# # dlm_refall_wrapper(i,ref_files)