/
2D signal-to-noise.py
134 lines (119 loc) · 4.45 KB
/
2D signal-to-noise.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
# coding: utf-8
#2-D signals
import numpy as np
import matplotlib.pyplot as plt
from cmath import polar
from PIL import Image
from random import randint
def create_shift(K):
l=[]
for i in range(K):
l.append((randint(0,40),randint(0,40)))
return l
#K=number of shift, sigma=noise magnitude
def signal_to_noise(K,noise_level):
image_file = Image.open("/Users/victorstorchan/Desktop/RA/ra/apple_original signal.png") # open colour image
image_signal = image_file.convert('L') # convert image to black and white
signal=np.asarray(image_signal)
(a,b)=signal.shape
randommat=np.zeros((400,400),dtype=complex)
for i in range(400):
for j in range(400):
randommat[i][j]+=complex(np.random.normal(0,noise_level,1))
signal_noisy=np.zeros((400,400),dtype=complex)
for i in range(a):
for j in range(b):
signal_noisy[10+i][10+j]=signal[i][j]
for i in range(400):
for j in range(400):
signal_noisy[i][j]+=randommat[i][j]
m=signal_noisy.shape[0]
n=signal_noisy.shape[1]
vect_of_shift=create_shift(K)
len_shift=len(vect_of_shift)
#signal with shift:
shifted_signals=[]
shifted_signals_1=[]
shifted_signals_2=[]
for s in range(len_shift):
y=np.zeros((n,m),dtype=complex)
y1=np.zeros((n,m),dtype=complex)
y2=np.zeros((n,m),dtype=complex)
for k in range(m):
for l in range(n):
if (l<10+vect_of_shift[s][0] or l>315+vect_of_shift[s][0]) and (k<10+vect_of_shift[s][1] or k>324+vect_of_shift[s][1]):
y[k][l]=randommat[k][l]
y1[k][l]=randommat[k][l]*np.exp(2J*np.pi*k/m)
y2[k][l]=randommat[k][l]*np.exp(2J*np.pi*l/n)
else:
y[k][l]=signal_noisy[k-vect_of_shift[s][0]][l-vect_of_shift[s][1]]
y1[k][l]=signal_noisy[k-vect_of_shift[s][0]][l-vect_of_shift[s][1]]*np.exp(2J*np.pi*k/m)
y2[k][l]=signal_noisy[k-vect_of_shift[s][0]][l-vect_of_shift[s][1]]*np.exp(2J*np.pi*l/n)
shifted_signals.append(y)
shifted_signals_1.append(y1)
shifted_signals_2.append(y2)
A=[]
A_1=[]
A_2=[]
for i in range(len(shifted_signals)):
A.append(np.matrix(np.fft.fft2(shifted_signals[i])))
A_1.append(np.matrix(np.fft.fft2(shifted_signals_1[i])).conjugate())
A_2.append(np.matrix(np.fft.fft2(shifted_signals_2[i])).conjugate())
G1=[]
G2=[]
for s in range(len_shift):
G1.append(np.multiply(A[s],A_1[s]))
G2.append(np.multiply(A[s],A_2[s]))
A_mat1=np.zeros((len_shift,n**2),dtype=complex)
A_mat2=np.zeros((len_shift,n**2),dtype=complex)
for s in range(len_shift):
for i in range(n):
for k in range(n):
A_mat1[s][400*i+k]=G1[s][i,k]
A_mat2[s][400*i+k]=G2[s][i,k]
A_mat1_mat=np.matrix(A_mat1)
A_mat2_mat=np.matrix(A_mat2)
A_mat1_transpose=A_mat1_mat.getH()
A_mat2_transpose=A_mat2_mat.getH()
A_prod1=A_mat1_mat*A_mat1_transpose
A_prod2=A_mat2_mat*A_mat2_transpose
A_final1=A_prod1/A_prod1[0,0]
A_final2=A_prod2/A_prod2[0,0]
(V1,sigma1,V_star1)=np.linalg.svd(A_final1)
(V2,sigma2,V_star2)=np.linalg.svd(A_final2)
v1=V_star1[0].getH()
v2=V_star2[0].getH()
#the shifts are recovered:
output1=np.zeros(len_shift)
output2=np.zeros(len_shift)
for i in range(len_shift):
output1[i]=-n*polar(-v1[i,0])[1]/(2*np.pi).real
output2[i]=-n*polar(-v2[i,0])[1]/(2*np.pi).real
recover_A=[]
for i in range(len_shift):
M=np.matrix(np.zeros((m,n),dtype=complex))
for l in range(m):
for k in range(n):
M[l,k]=A[i][l,k]/np.exp(-2J*np.pi*(l*output1[i]+k*output2[i])/n)
recover_A.append(M)
A_final=[]
for i in range(len_shift):
A_final.append(np.fft.ifft2(recover_A[i]))
recov_signal=np.zeros((m,n))
for i in range(m):
for j in range(n):
k=0
for s in range(len_shift):
k+=A_final[s][i][j].real
recov_signal[i][j]=k/len_shift
recov_signal1=recov_signal.astype(np.uint8)
return np.linalg.norm(recov_signal1-im3)
#k is just the count of the number of iteration to see how the program is running, because
#it is slow.
result_matrix=np.zeros((100,100))
k=0
for i in range(2,100):
for j in range(1,100):
result_matrix[99-i][j]=signal_to_noise(i,j)
k+=1
print(k)