-
Notifications
You must be signed in to change notification settings - Fork 1
/
full_ref_tech.py
80 lines (73 loc) · 2.93 KB
/
full_ref_tech.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
from __future__ import absolute_import, division, print_function
import numpy as np
import sys
import pandas as pd
from utils import _initial_check,_get_sigmas,_get_sums,Filter,fspecial,filter2
def _vif_single(Org,Dis,sigma_n):
"""
Add comment
"""
EPS = 1e-10 #tolernace for zero variance. Variance below this is set to zero,
#and zero is set to this value to avoid numerical issues such as multiplication by zero in some cases
num =0.0
den =0.0
for scale in range(1,5):
#will go from 1,2,3,4
# compute the size of the window used in the distortion channel estimation
N=2.0**(4-scale+1)+1
win = fspecial(Filter.GAUSSIAN,ws=N,sigma=N/5)
if scale >1:
Org = filter2(Org,win,'valid')[::2, ::2]
Dis = filter2(Dis,win,'valid')[::2, ::2]
#getting variances
Org_sum_sq,Dis_sum_sq,Org_Dis_sum_mul = _get_sums(Org,Dis,win,mode='valid')
#print(Org_sum_sq)
#getting covariances
sigmaOrg_sq,sigmaDis_sq,sigmaOrg_Dis = _get_sigmas(Org,Dis,win,mode='valid',sums=(Org_sum_sq,Dis_sum_sq,Org_Dis_sum_mul))
#get rid of numerical problems, very small negative numbers, or very
#small positive numbers, or other theoretical impossibilities.
sigmaOrg_sq[sigmaOrg_sq<0]=0
#print(sigmaGT_sq)
sigmaDis_sq[sigmaDis_sq<0]=0
#regression step to get values of g i.e.
g=sigmaOrg_Dis /(sigmaOrg_sq+EPS)
#Variance of error in regression
sv_sq=sigmaDis_sq-g*sigmaOrg_Dis
#get rid of numerical problems, very small negative numbers, or very
#small positive numbers, or other theoretical impossibilities.
g[sigmaOrg_sq<EPS]=0
sv_sq[sigmaOrg_sq<EPS]=sigmaDis_sq[sigmaOrg_sq<EPS]
sigmaOrg_sq[sigmaOrg_sq<EPS]=0
g[sigmaDis_sq<EPS]=0
sv_sq[sigmaDis_sq<EPS]=0
# constrain g to be non-negative.
sv_sq[g<0]=sigmaDis_sq[g<0]
g[g<0]=0
#take care of numerical errors, vv could be very small negative
sv_sq[sv_sq<=EPS]=EPS
#print(np.sum([[1,2],[3,4]]))
#denominator deontes reference image information (E)
#numerator denotes distorted image information (F)
# g= gi.......sigmaGT_sq= (Si^2 * Lamb(k))......sv_sq = sigma(v)^2.......sigma_n = sig(n)^2
num += np.sum(np.log10(1.0+(g**2.)*sigmaOrg_sq/(sv_sq+sigma_n)))
den += np.sum(np.log10(1.0+sigmaOrg_sq/sigma_n))
#print(num/den)
return num/den
def vif(Org,Dis,sigma_n=2):
"""calculates Pixel Based Visual Information Fidelity (vif).
Org: first (original) input image.
Dis: second (deformed) input image.
sigma_n: variance of the visual noise (default = 2)
"""
Org,Dis = _initial_check(Org,Dis)
#checking whther we are getting correct index within 0 and 1 or not
#print(Org.shape[2])
"""print(Org[:,:,0])
print(GT[:,:,1])
print(GT[:,:,2])
print(GT[:,:,3])
"""
#print(_vifp_single(GT[:,:,1],P[:,:,1],sigma_n))
# GT,P = GT[:,:,np.newaxis],P[:,:,np.newaxis]
#3 times because of r, g, b separate calculations
return np.mean([_vif_single(Org[:,:,i],Dis[:,:,i],sigma_n) for i in range(Org.shape[2])])