heatmap color RGB data for manaully created image

I have the usage data for they key use on a qwerty keyboard. I would like to create a heat map on the image of a traditional keyboard (made at the end).

I really like how heatmap works with the color scheme. How can I get the color information from the heatmap for my keyboard? How can I can I find the RGB values for each key?

I found similiar questions to this, but they were not clear.

% Given data: the sum of Data is equal to 1
% sum(data) = 1 is True
data = [0.000185322461082283        0.0209811500582442         0.121240601503759        0.0536905644392672        0.0794967171449751        0.0211333792227047       0.0234035793709626        0.0543855236683257        0.0746584771788626        0.0124761728264323;...
        0.0893254262416605        0.0582243460764588         0.0493354866038335        0.0234763846235307        0.0161164354548343        0.0785303928836175       0.0033556602774542       0.00686354971936884        0.0357407603515832        0.0146801863814466;...
      0.000781001800275336      0.000344170284867097         0.0167716827279466       0.00929921635073599      0.0168974372551096        0.0614145398708038        0.0265209679127396        0.0195912316001271       0.0101066927883088      0.000972942920681987];
layout =[  "q"    "w"    "e"    "r"    "t"    "y"    "u"    "i"    "o"    "p"
    "a"    "s"    "d"    "f"    "g"    "h"    "j"    "k"    "l"    ";"
    "z"    "x"    "c"    "v"    "b"    "n"    "m"    ","    "."    "/"];
% heatmap generate
figure(1)
fig = heatmap(data,'Colormap',parula)

fig = 
  HeatmapChart with properties:

        XData: {10×1 cell}
        YData: {3×1 cell}
    ColorData: [3×10 double]

  Show all properties

% values found for the making image, key size, key widths
key_num_width = 10.80;
key_num_hieght = 3;
key_pixel = 30;
size_multiplier = 10; % mess with this to change resolution probably something close to 10

% I forget what this does
A = 255*ones(size_multiplier*key_num_hieght*key_pixel,size_multiplier*key_num_width*key_pixel);
[B, C] = deal(A);
key_size = size_multiplier*key_pixel;
% changes the ones to 255 for image
rgbImage(:,:,1) = uint8(A);
rgbImage(:,:,2) = uint8(B);
rgbImage(:,:,3) = uint8(C);

Here, I need the data which are the three RGB matrixes each being 3x10

% HERE 
% I would like to find the RGB matrix values from the heatmap
R_value = 200*ones(3,10)

R_value = 3×10

200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

G_value = 200*ones(3,10)

G_value = 3×10

200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

B_value = 200*ones(3,10)

B_value = 3×10

200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200 200

% specify key colors given R, G and B matrixes (all 3x10) that has been
% extracted from heatmap
% starts top row (left to right) and then down each row one at a time
offset = [0 0.25 0.75]; % offset for key rows
for j = 0:2
    % specifies the boundries in the y directions for the keys
    ywidth = (key_size*j+1):(key_size+key_size*j);
    for i = 0:9
        % specifies the boundries in the x directions for the keys
        % note that there is a offset applied at 0, 0.25 and 0.75 of
        % keywidth
        xwidth = (key_size*i+1+key_size*offset(j+1)):(key_size+key_size*i+key_size*offset(j+1));
        % apply found RGB values from heat map to key at a time
        rgbImage(ywidth,xwidth,1) = R_value(j+1,i+1);
        rgbImage(ywidth,xwidth,2) = G_value(j+1,i+1);
        rgbImage(ywidth,xwidth,3) = B_value(j+1,i+1);
    end
end

% enter text for each key, included in question for clarity
RGB = rgbImage;
for i = 0:9
    RGB = insertText(RGB,[ key_size/2+i*key_size,key_size/2],layout(1,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end
for i = 0:9
    RGB = insertText(RGB,[key_size/2+i*key_size+key_size*0.25,key_size/2+key_size],layout(2,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end
for i = 0:9
    RGB = insertText(RGB,[key_size/2+i*key_size+0.75*key_size,key_size/2+key_size*2],layout(3,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end

% show result
figure(2)
imshow(RGB)

[Adam Danz]

Welcome to the forum @Taybs. I edited your question to produce the results of your sample code using the run feature (green triangle in the editor).

[Taybs]

Thank you Adam!

[Adam Danz]

This is the first time I see a heat map created with insertText! Ha! Creative.

Answer 1

[DGM]

The use of heatmap() is unnecessary. This is a basic use of indexed color. Generate a color table and find the indexes into it.

data = [0.000185322461082283 0.0209811500582442 0.121240601503759 0.0536905644392672 0.0794967171449751 0.0211333792227047 0.0234035793709626 0.0543855236683257 0.0746584771788626 0.0124761728264323;...
    0.0893254262416605 0.0582243460764588 0.0493354866038335 0.0234763846235307 0.0161164354548343 0.0785303928836175 0.0033556602774542 0.00686354971936884 0.0357407603515832 0.0146801863814466;...
    0.000781001800275336 0.000344170284867097 0.0167716827279466 0.00929921635073599 0.0168974372551096 0.0614145398708038 0.0265209679127396 0.0195912316001271 0.0101066927883088 0.000972942920681987];

ctlength = 256; % pick a color table length
CT = parula(ctlength); % generate a color table
ctidx = round(mat2gray(data)*(ctlength-1)) % these are the corresponding row indices into CT

ctidx = 3×10

0 44 255 113 167 44 49 114 157 26 188 122 104 49 34 165 7 14 75 31 1 0 35 19 35 129 55 41 21 2

What is the color associated with data(1,3)?

CT(ctidx(1,3)+1,:)

ans = 1×3

0.9769 0.9839 0.0805

All you need to do at this point is apply that color to the image regions.

[Taybs]

Thank you for you reply. How do you take the output

CT(ctidx(1,3)+1,:)

and get the RGB value from

0.9769 0.9839 0.0805

do I multiply by 255?

[DGM]

Here's a kludge using the existing code.

data = [0.000185322461082283 0.0209811500582442 0.121240601503759 0.0536905644392672 0.0794967171449751 0.0211333792227047 0.0234035793709626 0.0543855236683257 0.0746584771788626 0.0124761728264323;...
    0.0893254262416605 0.0582243460764588 0.0493354866038335 0.0234763846235307 0.0161164354548343 0.0785303928836175 0.0033556602774542 0.00686354971936884 0.0357407603515832 0.0146801863814466;...
    0.000781001800275336 0.000344170284867097 0.0167716827279466 0.00929921635073599 0.0168974372551096 0.0614145398708038 0.0265209679127396 0.0195912316001271 0.0101066927883088 0.000972942920681987];

layout =[  "q"    "w"    "e"    "r"    "t"    "y"    "u"    "i"    "o"    "p"
    "a"    "s"    "d"    "f"    "g"    "h"    "j"    "k"    "l"    ";"
    "z"    "x"    "c"    "v"    "b"    "n"    "m"    ","    "."    "/"];

ctlength = 256;
CT = im2uint8(parula(ctlength));
ctidx = round(mat2gray(data)*(ctlength-1));

% values found for the making image, key size, key widths
key_num_width = 10.80;
key_num_hieght = 3;
key_pixel = 30;
size_multiplier = 10; % mess with this to change resolution probably something close to 10

% preallocate image
rgbImage = 255*ones(size_multiplier*key_num_hieght*key_pixel,size_multiplier*key_num_width*key_pixel,3,'uint8');
key_size = size_multiplier*key_pixel;

% specify key colors given R, G and B matrixes (all 3x10) that has been
% extracted from heatmap
% starts top row (left to right) and then down each row one at a time
offset = [0 0.25 0.75]; % offset for key rows
for j = 0:2
    % specifies the boundries in the y directions for the keys
    ywidth = (key_size*j+1):(key_size+key_size*j);
    for i = 0:9
        % specifies the boundries in the x directions for the keys
        % note that there is a offset applied at 0, 0.25 and 0.75 of
        % keywidth
        xwidth = (key_size*i+1+key_size*offset(j+1)):(key_size+key_size*i+key_size*offset(j+1));
        % apply found RGB values from heat map to key at a time
        for c = 1:3
            rgbImage(ywidth,xwidth,c) = CT(ctidx(j+1,i+1)+1,c);
        end
    end
end

% enter text for each key, included in question for clarity
RGB = rgbImage;
for i = 0:9
    RGB = insertText(RGB,[key_size/2+i*key_size,key_size/2],layout(1,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
    RGB = insertText(RGB,[key_size/2+i*key_size+key_size*0.25,key_size/2+key_size],layout(2,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
    RGB = insertText(RGB,[key_size/2+i*key_size+0.75*key_size,key_size/2+key_size*2],layout(3,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end

% show result
imshow(RGB)

[Adam Danz]

Nice use of mat2gray :)

[DGM]

The name is camouflage to make sure nobody knows to use it, but I like that it doesn't have a bunch of varied options to remember. It's nice and simple.

[DGM]

FWIW, this is probably how I would have approached it with MIMT. It's definitely more costly in terms of system resources, but at least it's easier for me to think about the image manipulation this way. The row offset routine could be converted into a function to reduce repeated code.

data = [0.000185322461082283 0.0209811500582442 0.121240601503759 0.0536905644392672 0.0794967171449751 0.0211333792227047 0.0234035793709626 0.0543855236683257 0.0746584771788626 0.0124761728264323;...
    0.0893254262416605 0.0582243460764588 0.0493354866038335 0.0234763846235307 0.0161164354548343 0.0785303928836175 0.0033556602774542 0.00686354971936884 0.0357407603515832 0.0146801863814466;...
    0.000781001800275336 0.000344170284867097 0.0167716827279466 0.00929921635073599 0.0168974372551096 0.0614145398708038 0.0265209679127396 0.0195912316001271 0.0101066927883088 0.000972942920681987];

layout = 'qwertyuiopasdfghjkl;zxcvbnm,./';
losize = size(data); % key layout [rows cols]

keysize = [40 40]; % in pixels
rowoffset = [0 0.25 0.75]; % in keywidth units
padcolor = [1 1 1]; % the padding color where rows are offset
textcolor = [0 0 0]; % the text color obviously

ctlength = 256; % pick a color table length
CT = parula(ctlength); % pick a color table

% generate row indices into color table
ctidx = round(mat2gray(data)*(ctlength-1));
% create initial 3x10x3 RGB image from CT
keypict = ind2rgb(ctidx,CT);
% expand it based on selected key size
keypict = repelem(keypict,keysize(1),keysize(2));

% pad it as necessary for row offsets
keypict = addborder(keypict,[0 0 0 ceil(max(rowoffset)*keysize(2))],padcolor,'normalized');
% offset key rows
for r = 1:losize(1)
    yrange = (r-1)*keysize(1)+1:(r-1)*keysize(1)+keysize(1);
    keypict(yrange,:,:) = circshift(keypict(yrange,:,:),round(rowoffset(r)*keysize(2)),2);
end

% construct text images
keytext = cell(numel(layout),1);
for k = 1:numel(layout)
    % i don't have CVT or insertText(), so i'm using crispy old bitmap fonts
    keytext{k} = textim(layout(k),'cordata-ppc21');
end
% pad the text objects to the same size as the keys (implicitly centered)
keytext = imstacker(keytext,'padding',0,'size',keysize);
% tile into a flat image, pad and offset as before
keytext = imtile(keytext,losize);
keytext = addborder(keytext,[0 0 0 ceil(max(rowoffset)*keysize(2))],0);
for r = 1:losize(1)
    yrange = (r-1)*keysize(1)+1:(r-1)*keysize(1)+keysize(1);
    keytext(yrange,:,:) = circshift(keytext(yrange,:,:),round(rowoffset(r)*keysize(2)),2);
end

% composite the two images
outpict = replacepixels(textcolor,keypict,keytext);

imshow(outpict)

I know nobody really needs this, but I had it half-written and didn't feel like deleting it.

[Adam Danz]

@DGM, normalize(A,'range') is similar and doesn't require Image Processing TB but it currently doesn't have an option to specify both the "range" method and the dimensions.

B = rescale(A) is even better IMO.

Answer 2

[Adam Danz]

See the 5 lines of code under "new section". I didn't change anything else.

% Given data: the sum of Data is equal to 1
% sum(data) = 1 is True
data = [0.000185322461082283        0.0209811500582442         0.121240601503759        0.0536905644392672        0.0794967171449751        0.0211333792227047       0.0234035793709626        0.0543855236683257        0.0746584771788626        0.0124761728264323;...
        0.0893254262416605        0.0582243460764588         0.0493354866038335        0.0234763846235307        0.0161164354548343        0.0785303928836175       0.0033556602774542       0.00686354971936884        0.0357407603515832        0.0146801863814466;...
      0.000781001800275336      0.000344170284867097         0.0167716827279466       0.00929921635073599      0.0168974372551096        0.0614145398708038        0.0265209679127396        0.0195912316001271       0.0101066927883088      0.000972942920681987];
layout =[  "q"    "w"    "e"    "r"    "t"    "y"    "u"    "i"    "o"    "p"
    "a"    "s"    "d"    "f"    "g"    "h"    "j"    "k"    "l"    ";"
    "z"    "x"    "c"    "v"    "b"    "n"    "m"    ","    "."    "/"];
% heatmap generate
figure(1)
fig = heatmap(data,'Colormap',parula)

fig = 
  HeatmapChart with properties:

        XData: {10×1 cell}
        YData: {3×1 cell}
    ColorData: [3×10 double]

  Show all properties

% values found for the making image, key size, key widths
key_num_width = 10.80;
key_num_hieght = 3;
key_pixel = 30;
size_multiplier = 10; % mess with this to change resolution probably something close to 10

% I forget what this does
A = 255*ones(size_multiplier*key_num_hieght*key_pixel,size_multiplier*key_num_width*key_pixel);
[B, C] = deal(A);
key_size = size_multiplier*key_pixel;
% changes the ones to 255 for image
rgbImage(:,:,1) = uint8(A);
rgbImage(:,:,2) = uint8(B);
rgbImage(:,:,3) = uint8(C);

% HERE 
% % I would like to find the RGB matrix values from the heatmap
% R_value = 200*ones(3,10)
% G_value = 200*ones(3,10)
% B_value = 200*ones(3,10)

New section

cmap = uint8(parula(256) * 255);
cidx = round(normalize(data(:),'range')*(height(cmap)-1))+1;
R_value = reshape(cmap(cidx,1),size(data));
G_value = reshape(cmap(cidx,2),size(data));
B_value = reshape(cmap(cidx,3),size(data));

% specify key colors given R, G and B matrixes (all 3x10) that has been
% extracted from heatmap
% starts top row (left to right) and then down each row one at a time
offset = [0 0.25 0.75]; % offset for key rows
for j = 0:2
    % specifies the boundries in the y directions for the keys
    ywidth = (key_size*j+1):(key_size+key_size*j);
    for i = 0:9
        % specifies the boundries in the x directions for the keys
        % note that there is a offset applied at 0, 0.25 and 0.75 of
        % keywidth
        xwidth = (key_size*i+1+key_size*offset(j+1)):(key_size+key_size*i+key_size*offset(j+1));
        % apply found RGB values from heat map to key at a time
        rgbImage(ywidth,xwidth,1) = R_value(j+1,i+1);
        rgbImage(ywidth,xwidth,2) = G_value(j+1,i+1);
        rgbImage(ywidth,xwidth,3) = B_value(j+1,i+1);
    end
end

% enter text for each key, included in question for clarity
RGB = rgbImage;
for i = 0:9
    RGB = insertText(RGB,[ key_size/2+i*key_size,key_size/2],layout(1,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end
for i = 0:9
    RGB = insertText(RGB,[key_size/2+i*key_size+key_size*0.25,key_size/2+key_size],layout(2,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end
for i = 0:9
    RGB = insertText(RGB,[key_size/2+i*key_size+0.75*key_size,key_size/2+key_size*2],layout(3,i+1),'FontSize',150,'AnchorPoint','Center','BoxOpacity',0);
end

% show result
figure(2)
imshow(RGB)

[Taybs]

Thank you so much. I appreciate it. I saw the edits that you made as well.