Video_Sparse_Coding/ExtractSIFT.m at master · aniketde/Video_Sparse_Coding · GitHub

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
function [feaSet, lenStat] = ExtractSIFT(I, gridSpacing, patchSize, maxImSize, nrml_threshold)
%==========================================================================
% usage: calculate the sift descriptors given the image directory
%
% inputs
% rt_img_dir    -image database root path
% rt_data_dir   -feature database root path
% gridSpacing   -spacing for sampling dense descriptors
% patchSize     -patch size for extracting sift feature
% maxImSize     -maximum size of the input image
% nrml_threshold    -low contrast normalization threshold
%
% outputs
% database      -directory for the calculated sift features
%
% Lazebnik's SIFT code is used.
%
% written by Jianchao Yang
% Mar. 2009, IFP, UIUC
%==========================================================================


            if ndims(I) == 3,
                I = im2double(rgb2gray(I));
            else
                I = im2double(I);
            end;

            [im_h, im_w] = size(I);

            if max(im_h, im_w) > maxImSize,
                I = imresize(I, maxImSize/max(im_h, im_w), 'bicubic');
                [im_h, im_w] = size(I);
            end;

            % make grid sampling SIFT descriptors
            remX = mod(im_w-patchSize,gridSpacing);
            offsetX = floor(remX/2)+1;
            remY = mod(im_h-patchSize,gridSpacing);
            offsetY = floor(remY/2)+1;

            [gridX,gridY] = meshgrid(offsetX:gridSpacing:im_w-patchSize+1, offsetY:gridSpacing:im_h-patchSize+1);

%             fprintf('Processing %s: wid %d, hgt %d, grid size: %d x %d, %d patches\n', ...
%                      frames(jj).name, im_w, im_h, size(gridX, 2), size(gridX, 1), numel(gridX));

            % find SIFT descriptors
            siftArr = sp_find_sift_grid(I, gridX, gridY, patchSize, 0.8);
            [siftArr, siftlen] = sp_normalize_sift(siftArr, nrml_threshold);

%             siftLens = [siftLens; siftlen];


            feaSet.feaArr = siftArr';
            feaSet.x = gridX(:) + patchSize/2 - 0.5;
            feaSet.y = gridY(:) + patchSize/2 - 0.5;
            feaSet.width = im_w;
            feaSet.height = im_h;

%             [pdir, fname] = fileparts(frames(jj).name);
%             fpath = fullfile(rt_data_dir, subname, [fname, '.mat']);
%
%             save(fpath, 'feaSet');
%             database.path = [database.path, fpath];

% lenStat = hist(siftLens, 100);