forked from csjunxu/TWSC-ECCV2018
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathDemo_TWSC_Sigma_RW_DND2017.m
85 lines (82 loc) · 3.05 KB
/
Demo_TWSC_Sigma_RW_DND2017.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
%-------------------------------------------------------------------------------------------------------------
% This is an implementation of the TWSC algorithm for real-world image denoising
%
% Author: Jun Xu, [email protected] / [email protected]
% The Hong Kong Polytechnic University
%
% Please refer to the following paper if you find this code helps:
%
% @article{TWSC_ECCV2018,
% author = {Jun Xu and Lei Zhang and David Zhang},
% title = {A Trilateral Weighted Sparse Coding Scheme for Real-World Image Denoising},
% journal = {ECCV},
% year = {2018}
% }
%
% Please see the file License.txt for the license governing this code.
%-------------------------------------------------------------------------------------------------------------
clear;
Original_image_dir = '../dnd_2017/images_srgb/';
fpath = fullfile(Original_image_dir, '*.mat');
im_dir = dir(fpath);
im_num = length(im_dir);
load '../dnd_2017/info.mat';
method = 'TWSC';
dataset = 'dnd_2017';
% write image directory
write_MAT_dir = ['../' dataset '_Results/'];
write_sRGB_dir = [write_MAT_dir method];
if ~isdir(write_sRGB_dir)
mkdir(write_sRGB_dir)
end
% set parameters
Par.ps = 6; % patch size
Par.step = 3; % the step of two neighbor patches
Par.win = 20; % size of window around the patch
Par.Outerloop = 8; % iteration number of algorithm
Par.Innerloop = 2; % iteration number of block matching
Par.nlspini = 70; % initial number of patches
Par.display = 0; %
Par.delta = 0; %
Par.nlspgap = 0; % 10
Par.lambda1 = 0; % set this parameter positive to apply TWSC model
Par.lambda2 = 3; % parameter for estimating local noise level
alltime = zeros(im_num, 1, 'double');
for i = 1 :im_num
Par.image = i;
load(fullfile(Original_image_dir, im_dir(i).name));
S = regexp(im_dir(i).name, '\.', 'split');
[h,w,ch] = size(InoisySRGB);
% iterate over bounding boxes
Idenoised_crop_bbs = cell(1,20);
for j = 1:size(info(1).boundingboxes,1)
Par.nlsp = Par.nlspini; % number of non-local patches
IMinname = [S{1} '_' num2str(j)];
bb = info(i).boundingboxes(j,:);
Par.nim = InoisySRGB(bb(1):bb(3), bb(2):bb(4),:);
Par.I = Par.nim;
% noise estimation
for c = 1:ch
Par.nSig(c) = NoiseEstimation(Par.nim(:, :, c)*255, Par.ps)/255;
end
% initial PSNR and SSIM
fprintf('%s: \n', IMinname);
% denoising
t1=clock;
[IMout, Par] = TWSC_Sigma_RW(Par);
t2=clock;
% etime(t2,t1)
alltime(Par.image) = etime(t2, t1);
%% output
IMoutname = sprintf([write_sRGB_dir '/' method '_' dataset '_' IMinname '.png']);
imwrite(IMout, IMoutname);
Idenoised_crop_bbs{j} = single(IMout);
end
for j = 1:size(info(1).boundingboxes,1)
Idenoised_crop = Idenoised_crop_bbs{j};
save(fullfile(write_MAT_dir, sprintf('%04d_%02d.mat', i, j)), 'Idenoised_crop');
end
fprintf('Image %d/%d done\n', i,50);
end
% generate submission files
bundle_submission_srgb( write_MAT_dir );