MATLAB-Art/colour_SOM.m at master · NylePudding/MATLAB-Art · 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
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
function a = colour_SOM (n,d,ite)
% n = n by n size of grid
% d = training data
% ite = number of iterations

% a = n by n, trained colour map

a = rand(n*n,1,3);
figure
colours = size(d);
imshow(reshape(a,n,n,3))
learning_rate = linspace(1,0.01,ite);

for i = 1 : ite

    colours = size(d);

    random_ind = randi([1 colours(1)]);

    for j = 1 : n*n
        dists(j) = sqrt(sum(( d(random_ind,:) - a(j,:)) .^ 2));
    end

%     a1 = reshape(a,n,n,3);

    [~,ind] = sort(dists);

    x_c = floor(ind(1)/ n);
    y_c =  mod(ind(1),n);

    a(ind(1),:) = a(ind(1),:) + learning_rate(i)*( d(random_ind,:) - a(ind(1),:));

    x_e = x_c + 1;
    y_e = y_c;
    e_ind = n * y_e + x_e;

    if x_e <= n && x_e > 0
        a(e_ind,:) = a(e_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(e_ind,:));
    end

    x_w = x_c - 1;
    y_w = y_c;
    w_ind = n * y_w + x_e;

    if x_w <= n && x_w > 0
        a(w_ind,:) = a(w_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(w_ind,:));
    end

    x_n = x_c;
    y_n = y_c - 1;
    n_ind = n * y_n + x_n ;

    if y_n < n && y_n > 0
        a(n_ind,:) = a(n_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(n_ind,:));
    end

    x_s = x_c;
    y_s = y_c + 1;
    s_ind = n * y_s + x_s;

    if y_s < n && y_s > 0
        a(s_ind,:) = a(s_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(s_ind,:));
    end


    x_ne = x_c + 1;
    y_ne = y_c - 1;
    ne_ind = n * y_ne + x_ne;

    if x_ne <= n && x_ne > 0 && y_ne < n && y_ne > 0
        a(ne_ind,:) = a(ne_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(ne_ind,:));
    end

    x_nw = x_c - 1;
    y_nw = y_c - 1;
    nw_ind = n * y_nw + x_nw;

    if x_nw <= n && x_nw > 0 && y_nw < n && y_nw > 0
        a(nw_ind,:) = a(nw_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(nw_ind,:));
    end

    x_se = x_c + 1;
    y_se = y_c + 1;
    se_ind = n * y_se + x_se;

    if x_se <= n && x_se > 0 && y_se < n && y_se > 0
        a(se_ind,:) = a(se_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(se_ind,:));
    end

    x_sw = x_c - 1;
    y_sw = y_c + 1;
    sw_ind = n * y_sw + x_sw;

    if x_sw <= n && x_sw > 0 && y_sw < n && y_sw > 0
        a(sw_ind,:) = a(sw_ind,:) + (learning_rate(i)*learning_rate(i))*( d(random_ind,:) - a(sw_ind,:));
    end


%
%     for j = size
%
%     d = (sqrt((d(randi([1 size(d)])^2) + a(:,))


end