ComputerVision/hog.m at master · colleenpamchen/ComputerVision · 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
68
69
70
71
72
73
74
75
76
function ohist = hog(I)
%
% compute orientation histograms over 8x8 blocks of pixels
% orientations are binned into 9 possible bins
% I : grayscale image of dimension HxW
% ohist : orinetation histograms for each block. ohist is of dimension (H/8)x(W/8)x9

[h,w] = size(I); %size of the input
h2 = ceil(h/8); %the size of the output
w2 = ceil(w/8);
heightPadding = h2*8 - h;
widthPadding = w2*8 - w;

[initMag,initOri] = mygradient(I);
% [mag,ori] = mygradient(I);

paddedMag = padarray(initMag,[heightPadding widthPadding]);
mag = paddedMag(heightPadding+1:end,widthPadding+1:end);


paddedOri = padarray(initOri,[heightPadding widthPadding]);
ori = paddedOri(heightPadding+1:end,widthPadding+1:end);

thresh = 0.1*max(max(mag));  %threshold for edges

edgePixels = (mag>thresh);
nori = 9;       %number of orientation bins
binMins = (-pi/2) : pi/nori : (pi/2-pi/nori);
binMaxs = (-pi/2+pi/nori) : pi/nori : (pi/2);

% separate out pixels into orientation channels
ohist = zeros(h2,w2,nori);

for i = 1:nori

    minVal = binMins(i);
    maxVal = binMaxs(i);

  % create a binary image containing 1's for the pixels that are edges at this orientation
  B = edgePixels & ( ori >= minVal ) & ( ori <= maxVal);
  %   sum up the values over 8x8 pixel blocks.
    chblock = im2col(B,[8 8],'distinct');  %useful function for grabbing block

    for row = 0:h2-1
        for col = 0:w2-1

           rowStart = 8*row + 1;
           rowEnd = 8*row + 8;
           colStart = 8*col + 1;
           colEnd = 8*col + 8;

           %number of edges in this block at this orientation
           numOrientEdges = sum(sum(double( B(rowStart:rowEnd,colStart:colEnd) )));

           %number of edges total in this block
           numTotalEdges = sum(sum(double(edgePixels(rowStart:rowEnd,colStart:colEnd))));

           % sum over each block and store result in ohist
            %   ohist(:,:,i) = ...

           if(numTotalEdges < 1)
            ohist(row+1,col+1,i) = 0;
           else
%                ohist(row+1,col+1,i) = numOrientEdges/numTotalEdges;
                    ohist(row+1,col+1,i) = numOrientEdges;
           end

       end
    end
    % normalize the histogram so that sum over orientation bins is 1 for each block
    % NOTE: Don't divide by 0! If there are no edges in a block (ie. this counts sums to 0 for the block) then just leave all the values 0.

end


end