Get values x and y from a diagram according to a .png picture

I have a picture.png that I have downloaded on matlab (imread). This picture represents a diagram on which you can find the SEL according frequencies. I would like to extract the SEL values for a specific frequency: is it possible to do such a thing on matlab ? To get the value of the diagram (under a .png picture) into a matrix so I could extract the desired values.

NB: On this plot there are several lines of different colors.

UPDATE - Okay so I managed to have a working piece of code.Basically, I'm telling Matlab to recognise each lines' color on the .png picture in order to reproduce the plot in a .fig figure. But each line is superposed and it becomes difficult for Matlab for matching the colors.How can I do to extract these particular colors that are hidden behind an other line ?

CODE

% Blue color
decode(1).pppp1 = find(im1(:,:,1) == 0);
decode(1).pppp2 = find(im1(:,:,2) == 0);
decode(1).pppp3 = find(im1(:,:,3) == 255);
decode(1).RGB = [0 0 1];

% Red color
decode(2).pppp1 = find(im1(:,:,1) == 255);
decode(2).pppp2 = find(im1(:,:,2) == 0);
decode(2).pppp3 = find(im1(:,:,3) == 0);
decode(2).RGB = [1 0 0];

% Green color
decode(3).pppp1 = find(im1(:,:,1) == 63);
decode(3).pppp2 = find(im1(:,:,2) == 247);
decode(3).pppp3 = find(im1(:,:,3) == 100);
decode(3).RGB = [0 1 0];

XBornes = [1 5];
YBornes = [100 170];

x = logspace(XBornes(1),XBornes(2),size(im1,2));              
y = fliplr(linspace(YBornes(1),YBornes(2),size(im1,1)));

for index_color = 1:length(decode);
    dummy = intersect(decode(index_color).pppp1, decode(index_color).pppp2);
    dummy = intersect(dummy, decode(index_color).pppp3);
    bidon = im1(dummy);
    for index_intersect = 1:numel(dummy);
        [I,J] = ind2sub(size(im1),dummy(index_intersect));  
        %%Calcul des coordonnées en valeurs.
        pts(index_intersect).X = x(J);
        pts(index_intersect).Y = y(I);     
    end
    semilogx([pts.X],[pts.Y],'Color',decode(index_color).RGB)
    hold on
end

[Image Analyst]

Are you going to let us see the image? And what is "the SEL" and what is the decode structure?