forked from cuiyixin555/LMG
-
Notifications
You must be signed in to change notification settings - Fork 0
/
blind_deconv.m
211 lines (186 loc) · 4.97 KB
/
blind_deconv.m
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
function [kernel, interim_latent] = blind_deconv(y, lambda_dark, lambda_grad, opts)
%
% Do multi-scale blind deconvolution
%
%% Input:
% @y : input blurred image (grayscale);
% @lambda_dark: the weight for the L0 regularization on intensity
% @lambda_grad: the weight for the L0 regularization on gradient
% @opts: see the description in the file "demo_text_deblurring.m"
%% Output:
% @kernel: the estimated blur kernel
% @interim_latent: intermediate latent image
%
% The Code is created based on the method described in the following paper
% [1] Jinshan Pan, Deqing Sun, Hanspteter Pfister, and Ming-Hsuan Yang,
% Blind Image Deblurring Using Dark Channel Prior, CVPR, 2016.
% [2] Jinshan Pan, Zhe Hu, Zhixun Su, and Ming-Hsuan Yang,
% Deblurring Text Images via L0-Regularized Intensity and Gradient
% Prior, CVPR, 2014.
%
% Author: Jinshan Pan ([email protected])
% Date : 03/22/2016
% gamma correct
if opts.gamma_correct~=1
y = y.^opts.gamma_correct;
end
b = zeros(opts.kernel_size);
% set kernel size for coarsest level - must be odd
%minsize = max(3, 2*floor(((opts.kernel_size - 1)/16)) + 1);
%fprintf('Kernel size at coarsest level is %d\n', maxitr);
%%
ret = sqrt(0.5);
%%
maxitr=max(floor(log(5/min(opts.kernel_size))/log(ret)),0);
num_scales = maxitr + 1;
fprintf('Maximum iteration level is %d\n', num_scales);
%%
retv=ret.^[0:maxitr];
k1list=ceil(opts.kernel_size*retv);
k1list=k1list+(mod(k1list,2)==0);
k2list=ceil(opts.kernel_size*retv);
k2list=k2list+(mod(k2list,2)==0);
% derivative filters
dx = [-1 1; 0 0];
dy = [-1 0; 1 0];
% blind deconvolution - multiscale processing
for s = num_scales:-1:1
if (s == num_scales)
%%
% at coarsest level, initialize kernel
ks = init_kernel(k1list(s));
k1 = k1list(s);
k2 = k1; % always square kernel assumed
else
% upsample kernel from previous level to next finer level
k1 = k1list(s);
k2 = k1; % always square kernel assumed
% resize kernel from previous level
ks = resizeKer(ks,1/ret,k1list(s),k2list(s));
end;
%%%%%%%%%%%%%%%%%%%%%%%%%%%
cret=retv(s);
ys=downSmpImC(y,cret);
fprintf('Processing scale %d/%d; kernel size %dx%d; image size %dx%d\n', ...
s, num_scales, k1, k2, size(ys,1), size(ys,2));
%-----------------------------------------------------------%
%% Useless operation
if (s == num_scales)
[~, ~, threshold]= threshold_pxpy_v1(ys,max(size(ks)));
%% Initialize the parameter: ???
% if threshold<lambda_grad/10&&threshold~=0;
% % lambda_grad = threshold;
% % lambda_dark = threshold_image_v1(ys);
% % lambda_dark = lambda_grad;
% end
end
%-----------------------------------------------------------%
[ks, lambda_dark, lambda_grad, interim_latent] = blind_deconv_main(ys, ks, lambda_dark,...
lambda_grad, threshold, opts);
%% center the kernel
% temp = get_maxgra(interim_latent,floor(size(ks,1)*1.5));
% name = strcat('lmg',num2str(s),'.png');
% imwrite(temp/2,name);
ks = adjust_psf_center(ks);
ks(ks(:)<0) = 0;
sumk = sum(ks(:));
ks = ks./sumk;
%% set elements below threshold to 0
if (s == 1)
kernel = ks;
if opts.k_thresh>0
kernel(kernel(:) < max(kernel(:))/opts.k_thresh) = 0;
else
kernel(kernel(:) < 0) = 0;
end
kernel = kernel / sum(kernel(:));
end;
end;
%% end kernel estimation
end
%% Sub-function
function [k] = init_kernel(minsize)
k = zeros(minsize, minsize);
k((minsize - 1)/2, (minsize - 1)/2:(minsize - 1)/2+1) = 1/2;
end
%%
function sI=downSmpImC(I,ret)
%% refer to Levin's code
if (ret==1)
sI=I;
return
end
%%%%%%%%%%%%%%%%%%%
sig=1/pi*ret;
g0=[-50:50]*2*pi;
sf=exp(-0.5*g0.^2*sig^2);
sf=sf/sum(sf);
csf=cumsum(sf);
csf=min(csf,csf(end:-1:1));
ii=find(csf>0.05);
sf=sf(ii);
sum(sf);
I=conv2(sf,sf',I,'valid');
[gx,gy]=meshgrid([1:1/ret:size(I,2)],[1:1/ret:size(I,1)]);
sI=interp2(I,gx,gy,'bilinear');
end
%%
function k=resizeKer(k,ret,k1,k2)
%%
% levin's code
k=imresize(k,ret);
k=max(k,0);
k=fixsize(k,k1,k2);
if max(k(:))>0
k=k/sum(k(:));
end
end
%%
function nf=fixsize(f,nk1,nk2)
[k1,k2]=size(f);
while((k1~=nk1)|(k2~=nk2))
if (k1>nk1)
s=sum(f,2);
if (s(1)<s(end))
f=f(2:end,:);
else
f=f(1:end-1,:);
end
end
if (k1<nk1)
s=sum(f,2);
if (s(1)<s(end))
tf=zeros(k1+1,size(f,2));
tf(1:k1,:)=f;
f=tf;
else
tf=zeros(k1+1,size(f,2));
tf(2:k1+1,:)=f;
f=tf;
end
end
if (k2>nk2)
s=sum(f,1);
if (s(1)<s(end))
f=f(:,2:end);
else
f=f(:,1:end-1);
end
end
if (k2<nk2)
s=sum(f,1);
if (s(1)<s(end))
tf=zeros(size(f,1),k2+1);
tf(:,1:k2)=f;
f=tf;
else
tf=zeros(size(f,1),k2+1);
tf(:,2:k2+1)=f;
f=tf;
end
end
[k1,k2]=size(f);
end
nf=f;
end
%%