-
Notifications
You must be signed in to change notification settings - Fork 0
/
IKDE_BSDS500_searchN.m
170 lines (151 loc) · 6.36 KB
/
IKDE_BSDS500_searchN.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
clc;clear ; close all;
%% set path
addpath 'others'
addpath 'evals'
addpath 'OLRSC'
addpath 'SSC'
addpath 'functions'
addpath 'msseg'
addpath 'APC'
addpath 'Graph_based_segment'
%% set parameters for bipartite graph
para.alpha = 0.001; % affinity between pixels and superpixels
para.beta = 20; % scale factor in superpixel affinity
para.nb = 1; % number of neighbors for superpixels
para.rho = 1;
para.epochs = 20;
para.d = 50;
para.Nimgs = 500; % number of images in BSDS300/500
%% read numbers of segments used in the paper
bsdsRoot ='./database/BSDS500';
fid = fopen(fullfile('Nsegs_500.txt'),'r');
[BSDS_INFO] = fscanf(fid,'%d %d \n',[2,para.Nimgs]);
fclose(fid);
%% PRI,VoI,GCE,BDE.
PRI_all = zeros(para.Nimgs,1);
VoI_all = zeros(para.Nimgs,1);
GCE_all = zeros(para.Nimgs,1);
BDE_all = zeros(para.Nimgs,1);
%% Settings
% setup
nGCluster = 3; % number of subjects.
% dimension reduction
reduceDimension = @(data) dimReduction_PCA(data, 0);
% normalization
normalizeColumn = @(data) cnormalize_inplace(data);
% representation
buildRepresentation = @(data) OMP_mat_func(data, 3, 1e-6); % second parameter is sparsity
% spectral clustering
genLabel = @(affinity, nCluster) SpectralClustering(affinity, nCluster, 'Eig_Solver', 'eigs');
%%
Nseg_save = [];
for idxI =1:para.Nimgs
% read number of segments
tic;Nseg = BSDS_INFO(2,idxI);
% locate image
img_name = int2str(BSDS_INFO(1,idxI));
img_loc = fullfile(bsdsRoot,'images','test',[img_name,'.jpg']);
present = 'test';
if ~exist(img_loc,'file')
img_loc = fullfile(bsdsRoot,'images','train',[img_name,'.jpg']);
present = 'train';
if ~exist(img_loc,'file')
img_loc = fullfile(bsdsRoot,'images','val',[img_name,'.jpg']);
present = 'val';
end
end
img = im2double(imread(img_loc)); [X,Y,~] = size(img);
out_path = fullfile('results_searchN_denoise_n','BSDS500','ikde_1_1_APC_3_5',img_name);
if ~exist(out_path,'dir'), mkdir(out_path); end
% generate superpixels
[para_MS, para_FH] = set_parameters_oversegmentation(img_loc);
[seg,labels_img,seg_vals,seg_lab_vals,seg_edges,seg_img] = make_superpixels(img_loc,para_MS,para_FH);
%% construct graph
Np = X*Y; Nsp = 0;
for k = 1:length(seg), Nsp = Nsp + size(seg{k},2); end
W_Y = sparse(Nsp,Nsp); edgesXY = []; j = 1;
for k = 1:length(seg)
% for each over-segmentation
feature = seg_lab_vals{k};
feature = ikde(feature,1,1);
tmp1 = reduceDimension(feature);
tmp1 = normalizeColumn(tmp1);
R = buildRepresentation(tmp1');
R(1:length(feature)+1:end) = 0;
A = abs(R) + abs(R)';
nGCluster(idxI,k) = APclustering_w_o_ikde(feature);
index_tmp = genLabel(A, nGCluster(idxI,k));
% superpixel division
local_nodes = find(index_tmp == mode(index_tmp));
global_nodes = find(index_tmp ~= mode(index_tmp));
% first we construct the adjacent graph over all nodes
w = makeweights(seg_edges{k},feature,para.beta);
W_local = adjacency(seg_edges{k},w);
% assign local graph entries to fused new graph W, we will
% replace the nodes belongs to globla_nodes with value of
% global graph value
%W=zeros(size(feature,1),size(feature,1));
W=W_local;
% randomly generate two set of supperpxiels from Medium set
% p = length(global_nodes);
p = global_nodes;
% please choose different kinds of global graph combination
W_OLRSC = OLRSCGRAPH_n(feature,para);
W = assignGraphValue(W,W_OLRSC,p);
W = sparse(W);
Nk = size(seg{k},2); % number of superpixels in over-segmentation k
W_Y(j:j+Nk-1,j:j+Nk-1) = prune_knn(W,para.nb);
% affinities between pixels and superpixels
for i = 1:Nk
idxp = seg{k}{i}; % pixel indices in superpixel i
Nki = length(idxp);
idxsp = j + zeros(Nki,1);
edgesXY = [edgesXY; [idxp, idxsp]];
j = j + 1;
end
end
W_XY = sparse(edgesXY(:,1),edgesXY(:,2),para.alpha,Np,Nsp);
% affinity between a superpixel and itself is set to be the maximum 1.
W_Y(1:Nsp+1:end) = 1; B = [W_XY;W_Y];
%% Transfer cut
out_path_gt= fullfile('results_searchN_denoise_n','BSDS500','ikde_1_1_APC_3_5',img_name);
if ~exist(out_path_gt,'dir'), mkdir(out_path_gt); end
[gt_imgs, gt_cnt] = view_gt_segmentation(bsdsRoot,img,present,out_path_gt,img_name,para,0);
nclusters=1:40; E=zeros(length(nclusters),5);segs=cell(1,length(nclusters));
for ncluster=1:length(nclusters)
label_img = Tcut(B,nclusters(ncluster),[X,Y]);
% display the result
view_segmentation(img,label_img(:),out_path,img_name,0);
% Evaluation and save result
out_vals = eval_segmentation(label_img,gt_imgs);
E(ncluster,:)=[nclusters(ncluster),out_vals.PRI,out_vals.VoI,out_vals.GCE,out_vals.BDE];
segs{ncluster}=uint16(label_img);
end
out_seg_path = fullfile('results_searchN_denoise_n','BSDS500','ikde_1_1_APC_3_5','segs');
if ~exist(out_seg_path,'dir'), mkdir(out_seg_path); end
out_seg = fullfile(out_seg_path,[img_name, '.mat']);
save('-v7',out_seg, 'segs');
outname = fullfile(out_path,[img_name, '.mat']);
save('-v7',outname, 'E');
% load(outname);
% Evaluation and save result
[maxE,idx] = max(E(:,2));ti = toc;
fprintf('%6d %6s: %2d %9.6f, %9.6f, %9.6f, %9.6f %.4fs\n', idxI,...
img_name,E(idx,1), E(idx,2), E(idx,3), E(idx,4), E(idx,5),ti);
Neg_all(idxI) = E(idx,1);
PRI_all(idxI) = E(idx,2); VoI_all(idxI) = E(idx,3);
GCE_all(idxI) = E(idx,4); BDE_all(idxI) = E(idx,5);
end
fprintf('Mean: %14.6f, %9.6f, %9.6f, %9.6f \n', mean(PRI_all), mean(VoI_all), mean(GCE_all), mean(BDE_all));
fid_out = fopen(fullfile('results_searchN_denoise_n','BSDS500','ikde_1_1_APC_3_5','evaluation.txt'),'w');
for idxI=1:para.Nimgs
fprintf(fid_out,'%6d %9.6f, %9.6f, %9.6f, %9.6f \n', BSDS_INFO(1,idxI),...
PRI_all(idxI), VoI_all(idxI), GCE_all(idxI), BDE_all(idxI));
end
fprintf(fid_out,'Mean: %10.6f, %9.6f, %9.6f, %9.6f \n', mean(PRI_all), mean(VoI_all), mean(GCE_all), mean(BDE_all));
fclose(fid_out);
fid_out2 = fopen(fullfile('results_searchN_denoise_n','BSDS500','ikde_1_1_APC_3_5','Nsegs.txt'),'w');
for idxI=1:para.Nimgs
fprintf(fid_out2,'%6d %d \n', BSDS_INFO(1,idxI),Neg_all(idxI));
end
fclose(fid_out2);