-
Notifications
You must be signed in to change notification settings - Fork 0
/
demo_DSP_PMF_all2.m
268 lines (228 loc) · 12.7 KB
/
demo_DSP_PMF_all2.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
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
close all;
clear all;
% for CAL
%{
load('rand_img.mat')
category_all = {'airplanes','chair','crayfish', 'cup', ...
'dalmatian', 'dolphin', 'helicopter', 'flamingo', 'minaret', ...
'panda', 'rooster', 'schooner', 'snoopy', 'soccer_ball', ...
'sunflower'};
%}
%{
{'airplanes',[2,3; 14,15; 25,28; 43,48; 70,75; 92,96; 12,56; 124,159; 314,361; 652,746]}
{'beaver',[4,3; 22,23; 35,36; 1,15; 34,44; 4,34; 34,35; 27,28; 5,17; 13,19]}
{'cup',[15,14; 47,56; 25,33; 27,30; 47,52; 16,15; 21,25; 57,48; 39,40; 56,28]}
{'schooner',[13,19; 27,29; 30,59; 12,13; 16,19; 36,48; 44,52; 55,41; 52,49; 30,24]};
{'snoopy',[16,17; 10,14; 4,9; 8,7; 27,25; 35,21; 15,1; 26,27; 24,11; 7,25]};
{'sunflower',[5,11; 57,59; 41,2; 28,34; 32,31; 73,74; 26,34; 23,36; 21,22; 7,5]}
{'butterfly',[8,3; 27,30; 35,32; 31,49; 1,58; 66,67; 73,81; 3,15; 61,86; 16,19]}
{'minaret',[24,25; 8,16; 20,25; 44,45; 50,55; 51,42; 33,32; 72,73; 75,69; 60,70]};
{'rooster',[27,7; 26,28; 46,43; 6,14; 22,29; 36,31; 47,29; 44,39; 4,9; 16,2]};
{'lamp',[13,12; 25,28; 56,60; 13,23; 36,53; 49,52; 17,5; 60,52; 23,20; 44,6]};
{'dalmatian',[1,2; 14,26; 39,40; 46,48; 59,51; 14,7; 35,28; 46,44; 38,29; 28,16]};
{'helicopter',[6,4; 24,23; 34,28; 37,43; 42,49; 60,53; 68,71; 87,65; 82,53; 51,45]};
{'ketch',[103,111; 83,79; 80,55; 35,56; 11,5; 17,20; 56,51; 53,45; 99,102; 62,61]};
{'water_lilly',[1,16; 35,1; 26,27; 10,18; 11,16; 8,10; 15,19; 32,33; 2,5; 13,14]};
{'umbrella',[9,7; 25,22; 31,33; 60,73; 43,69; 45,61; 37,26; 26,40; 52,63; 50,54]};
{'ferry',[19,33; 40,41; 11,42; 58,67; 50,40; 33,44; 19,40; 48,31; 27,39; 4,14]};
{'flamingo',[7,6; 28,27; 10,28; 54,32; 67,62; 59,43; 56,51; 1,51; 9,45; 11,22]};
{'chair',[1,23; 24,29; 46,49; 36,37; 35,39; 16,9; 45,27; 31,33; 2,61; 14,4]};
{'panda',[3,4; 9,14; 11,31; 22,29; 29,30; 36,11; 7,13; 23,29; 4,10; 31,32]};
{'crayfish',[7,2; 16,31; 29,17; 43,40; 69,43; 47,57; 64,58; 16,14; 54,39; 59,47]};
%}
class_name_all = {{'helicopter',[6,4; 24,23; 34,28; 37,43; 42,49; 60,53; 68,71; 87,65; 82,53; 51,45]};
{'ketch',[103,111; 83,79; 80,55; 35,56; 11,5; 17,20; 56,51; 53,45; 99,102; 62,61]}};
%class_name_all = {{'rooster',27,7},{'rooster',26,28},{'rooster',46,43},...
% {'beaver',23,3}};
%{
{{'rooster',27,7},{'rooster',26,28},{'rooster',46,43},...
{'beaver',23,3},{'beaver',22,23},{'beaver',36,44},...
{'schooner',30,59},{'schooner',12,13},{'schooner',16,19}...
{'sunflower',41,2},{'sunflower',5,11},{'sunflower',57,59},...
{'cup',56,28},{'cup',47,51},{'cup',39,40},...
{'minaret',52,15},{'minaret',24,25},{'minaret',72,73}};
%}
% LMO
%category_all = {'mountain', 'street', 'tallbuilding'};%, 'coast_sun', 'highway','highway_car', 'insidecity', 'mountain', 'street', 'tallbuilding'};
% for VGG
%class_name_all = {'boat'};
%file_name1_all = {'_2'};%{'_2','_3','_4','_5','_6'};
fusion=1; %1:SGDL, 0:single descriptor
dataset='CAL'; %'LMO','VGG', 'CAL'
sift = [5 8 10 12 15 18 20];
gb = [3 4 5 6 7 8 9];
daisy = [10 15 20 25 30 35 40];
liop = [10 15 20 25 30 35 40];
for num_class=1:length(class_name_all)
for img_num = 1:size(class_name_all{num_class}{2},1) % for CAL: 1:10; for VGG 1:5; for LMO 1:9
for p_d = 3
if strcmp(dataset,'CAL')
%class_name = {category_all{class},rand_all.(category_all{class})(2*img_num-1),rand_all.(category_all{class})(2*img_num)};
class_name = class_name_all{num_class};
category = class_name{1};
ind1 = class_name{2}(img_num,1);
ind2 = class_name{2}(img_num,2);
name = [category,num2str(ind2),'to',num2str(ind1)];
input_dir = ['datasets\cal101\101_ObjectCategories\',category,'\'];
output_dir = ['Results\CAL\',category,'\',num2str(ind2),'to',num2str(ind1),'\'];
mkdir(output_dir);
output_dir = ['Results\CAL\',category,'\',num2str(ind2),'to',num2str(ind1), '\'];
output_dir2 = ['Results\CAL\',category,'\',num2str(ind2),'to',num2str(ind1), '\other\'];
output_dir3 = [input_dir name, '\supportR' num2str(10*(p_d+1)) '\'];
mkdir(output_dir);
mkdir(output_dir2);
mkdir(output_dir3);
if ind1<10
ind1 = ['000',num2str(ind1)];
elseif ind1<100
ind1 = ['00',num2str(ind1)];
else
ind1 = ['0',num2str(ind1)];
end
if ind2<10
ind2 = ['000',num2str(ind2)];
elseif ind2<100
ind2 = ['00',num2str(ind2)];
else
ind2 = ['0',num2str(ind2)];
end
im1=imread([input_dir,'image_',ind1,'.jpg']);
im2=imread([input_dir,'image_',ind2,'.jpg']);
elseif strcmp(dataset,'VGG')
class_name =class_name_all{num_class};
file_name1 = file_name1_all{img_num};
file_name2 = '_1';
input_dir = ['datasets\VGG\' class_name '\'];
im1=imread([input_dir 'img' file_name1(end) '.ppm']);
im2=imread([input_dir 'img' file_name2(end) '.ppm']);
name = [class_name,' ',file_name2(2:end) 'to' file_name1(2:end)];
output_dir = ['Results\VGG\' class_name '_test\' file_name2(2:end) 'to' file_name1(2:end) '\supportR' num2str(10*(p_d+1)) '\'];
output_dir2 = ['Results\VGG\' class_name '_test\' file_name2(2:end) 'to' file_name1(2:end) '\supportR' num2str(10*(p_d+1)) '\other\'];
output_dir3 = [input_dir name, '\supportR' num2str(10*(p_d+1)) '\'];
mkdir(output_dir);
mkdir(output_dir2);
mkdir(output_dir3);
%im1 = imresize(imfilter(im1,fspecial('gaussian',5,0.67),'same','replicate'),0.5,'bicubic');
%im2 = imresize(imfilter(im2,fspecial('gaussian',5,0.67),'same','replicate'),0.5,'bicubic');
end
%% preprocessing
if size(im1,3)<3
new_im=uint8(zeros(size(im1,1),size(im1,2),3));
new_im(:,:,1)=im1;
new_im(:,:,2)=im1;
new_im(:,:,3)=im1;
im1=new_im;
end
if size(im2,3)<3
new_im=uint8(zeros(size(im2,1),size(im2,2),3));
new_im(:,:,1)=im2;
new_im(:,:,2)=im2;
new_im(:,:,3)=im2;
im2=new_im;
end
% Set parameters
K = 500; % number of superpixels for one image
r = 12; % r-pixel for extended subimage
% tic;
[im1_seg, im1_graph]=SegImgSLIC(im1, K, r);
[im2_seg, im2_graph]=SegImgSLIC(im2, K, r);
% toc;
% Extract features
paras_d.SIFT = sift(p_d);
paras_d.GB = gb(p_d);
paras_d.DAISY = daisy(p_d);
paras_d.LIOP = liop(p_d);
%descps1=ExtractAllDescps(im1,paras_d);
%descps2=ExtractAllDescps(im2,paras_d);
% load('SIFT_01_a_descps.mat');
% load('SIFT_01_b_descps.mat');
%save([output_dir 'descp.mat'],'descps1', 'descps2','-v7.3');
if exist([output_dir3 'descp.mat'],'file');
load([output_dir3 'descp.mat']);
else
descps1=ExtractAllDescps(im1,paras_d);
descps2=ExtractAllDescps(im2,paras_d);
%save([output_dir3 'descp.mat'],'descps1', 'descps2','-v7.3');
end
descps1.im=im1;
descps2.im=im2;
im1=im2double(im1);
im2=im2double(im2);
%% patchmatch filter algorithm
% Set parameters
feature_num=4;
iter_times=20;
% alphas=[0.25 0.5 0.75 1 1.25 1.5]';
% alphas=0.7;
% winsizes=[149 99 49 39 15]';
% gamma2s=[1 2 3 4 5 6 7 8 9 10 20 30 40]';
dsp_gammas=[0.75];%0.25; %
dsp_alphas=[0.2];%0.1; %
% gamma1s=[0.5 0.6 0.7 0.8 0.9 1]';
% gamma1s=1;
% label_result=mexPatchMatchFilter(im1_graph,im2_graph);
Descps_type = {'SIFT','GB','DAISY','LIOP'};%{'SIFT','GB','DAISY','LIOP'};
for i=1:length(dsp_gammas)
for j=1:length(dsp_alphas)
if fusion==1
out_filename=['SGDL_alpha=',num2str(dsp_alphas(j)),',gamma=',num2str(dsp_gammas(i)), ',R=' num2str(10*(p_d+1))];%Descps_type{desp};%
% parameters setting
paras.dsp_alpha=dsp_alphas(j);
paras.dsp_gamma=dsp_gammas(i);
paras.alpha=0.7; % weighted value
paras.gamma1=1; % <= 1
paras.gamma2=3; % >= 1
paras.lambda=0.0000000001; % avoid to be divided by zero
paras.descps_type = Descps_type;%{Descps_type{desp}};%
paras.K=2;
if exist([output_dir2 out_filename '_label.mat'],'file');
load([output_dir2 out_filename '_label.mat']);
else
[dsp_label_result] = DSPMatch(descps1,descps2,paras);
vy=dsp_label_result(:,:,1);
vx=dsp_label_result(:,:,2);
warp21=warpImage(im2,vx,vy);
figure, imshow(warp21);
save([output_dir2 out_filename '_label.mat'],'dsp_label_result');
imwrite(warp21,[output_dir2 out_filename '_warp.png'],'png');
end
[label_result cost_result it]=DSP_PatchMatchFilter(im1_graph,im2_graph,descps1,descps2,dsp_label_result,feature_num,iter_times,paras, output_dir, out_filename);
vy=label_result(:,:,1);
vx=label_result(:,:,2);
desc_map=label_result(:,:,3);
cost=cost_result;
mkdir([output_dir 'comparison\']);
save([output_dir 'comparison\' out_filename '-iter=' num2str(it) '.mat'],'vx','vy','desc_map','paras','cost');
else
for desp = 1:length(Descps_type)
out_filename=[Descps_type{desp} '_alpha=',num2str(dsp_alphas(j)),',gamma=',num2str(dsp_gammas(i))];%Descps_type{desp};%
% parameters setting
paras.dsp_alpha=dsp_alphas(j);
paras.dsp_gamma=dsp_gammas(i);
paras.alpha=0.7; % weighted value
paras.gamma1=1; % <= 1
paras.gamma2=3; % >= 1
paras.lambda=0.0000000001; % avoid to be divided by zero
paras.descps_type = {Descps_type{desp}};
paras.K=2;
[dsp_label_result] = DSPMatch(descps1,descps2,paras);
vy=dsp_label_result(:,:,1);
vx=dsp_label_result(:,:,2);
warp21=warpImage(im2,vx,vy);
figure, imshow(warp21);
save([output_dir2 out_filename '_label.mat'],'dsp_label_result');
imwrite(warp21,[output_dir2 out_filename '_warp.png'],'png');
[label_result cost_result it time_all]=DSP_PatchMatchFilter(im1_graph,im2_graph,descps1,descps2,dsp_label_result,feature_num,iter_times,paras, output_dir, out_filename);
vy=label_result(:,:,1);
vx=label_result(:,:,2);
desc_map=label_result(:,:,3);
cost=cost_result;
mkdir([output_dir 'comparison\']);
save([output_dir 'comparison\' out_filename '-iter=' num2str(it) '.mat'],'vx','vy','desc_map','paras','cost','time_all');
end
end
end
end
end
end
end