imageRGB_write_file.m

%6.111 Image Color Table MATLAB deme
%Edgar Twigg bwayr@mit.edu
%4/1/2008 (But I swear this file isn't a joke)
close all
Number = 'raindrop_Pixel';

%% How to use this file
%Notice how %% divides up sections?  If you hit ctrl+enter, then MATLAB 
%will execute all the lines within that section, but nothing else.  You can
%also navigate quickly through the file using ctrl+arrow_key

%% Getting 24 bit data
%So when you look at a 24 bit bitmap file, the file specifies three 8 bit 
%values for each color, 8 each for red, green, and blue.
[picture] = imread(strcat('./Images/',Number,'.bmp'));

%% View the image
%This command image will draw the picture you just loaded
figure                  %opens a new window
image(picture)          %draws your picture
title('24 bit bitmap')  %gives it a title so you don't forget what it is

%% Manipulate the data in the image
%So now you have a matrix of values that represent the image.  You can
%access them in the following way:
%
%picture(row,column,color)
%
%Remember that MATLAB uses 1-based indexes, and Verilog uses 0!
%
%Also, you can use MATLAB's slice operator to do nifty things.
%picture(:,:,1) would return a 2D matrix with the red value for every row and
%column.
%
%picture(:,1,2) would return a 1D matrix with the green value for every row
%in the first column.

%This is how MATLAB indexes the colors
RED = 1;
GREEN = 2;
BLUE = 3;

%So if we wanted to see the red values of the image only, we could say
figure
image(picture(:,:,RED))
title('Red values in 24 bit bitmap')

%Because the image we gave matlab above specifies only one value per pixel
%rather than usual three (red,blue,green), MATLAB colors each pixel from
%blue to red based on the value at that pixel.

%% Getting 8 bit data
%When you store an 8 bit bitmap, things get a little more complicated.  Now
%each pixel in the image only gets one 8 bit value.  But, you need to send
%the monitor an r,g, and b!  How can this work?
%
%8 bit bitmaps include a table which specifies the rgb values for each of
%the 8 bits in the image.
%
%So each pixel is represented by one byte, and that byte is an index into a
%table where each index specifies an r, g, and b value separately.
%
%Because of this, now we need to load both the image and it's colormap.
[picture, color_table] = imread(strcat('./Images/',Number,'.bmp'));

%% Displaying without the color table
%If we try to display the picture without the colormap, the image does not
%make sense
figure
image(picture)
title('Per pixel values in 8 bit bitmap')

%% Displaying WITH the color table
%So to display the picture with the proper color table, we need to tell 
%MATLAB to set its colormap to be in line with our colorbar.  The image
%quality is somewhat reduced compared to the 24 bit image, but not too bad.
figure
image(picture)
colormap(color_table)   %This command tells MATLAB to use the image's color table
colorbar                %This command tells MATLAB to draw the color table it is using
title('8 bit bitmap displayed using color table')

% %% More about the color table
% %The color table is in the format:
% %
% %color_table(color_index,1=r 2=g 3=b)
% %
% %So to get the r g b values for color index 3, we only need to say:
% disp('         r    g              b    for color 3 is:')
% disp(color_table(3,:))      %disp = print to console
% 
% %Although in the bitmap file the colors are indexed as 0-255 and each rgb
% %value is an integer between 0-255, MATLAB images don't work like that, so
% %MATLAB has automatically scaled them to be indexed 1-256 and to have a
% %floating point value between 0 and 1.  To turn the floats into integer 
% %values between 0 and 256:
% 
% color_table_8bit = uint8(round(256*color_table));
% 
% disp('    r    g    b    for color 3 in integers is:')
% disp(color_table_8bit(3,:)) 
% 
% %Note that this doesn't fix the indexing (and it can't, since MATLAB won't
% %let you have indexes below 1)
% 
% %another way to look at the color table is like this (don't worry about how
% %to make this graph)
% figure
% stem3(color_table_8bit)
% set(gca,'XTick',1:3);
% set(gca,'YTick',[1,65,129,193,256]);
% set(gca,'YTickLabel',['  0';' 64';'128';'192';'255']);
% set(gca,'ZTick',[0,64,128,192,255]);
% 
% xlabel('red = 1, green = 2, blue = 3')
% ylabel('color index')
% zlabel('value')
% title('Another way to see the color table')

%% Even smaller bitmaps
%You can extend what we did for 8-bit bitmaps to even more compressed 
%forms, such as this 4-bit bitmap.  Now we only have 16 colors to work with
%though, and our image quality is significantly reduced:
% [picture, color_table] = imread('./Images/3.bmp');
% 
% figure
% image(picture)
% colormap(color_table)
% colorbar                                  
% title('4 bit bitmap displayed using color table')

%% Writing data to coe files for putting them on the fpga
%You can instantiate BRAMs to take their values from a file you feed them
%when you flash the FPGA.  You can use this technique to send them
%colortables, image data, anything.  Here's how to send the red component
%of the color table of the last example

red = color_table(:,1);             %grabs the red part of the colortable
scaled_data = red*255;              %scales the floats back to 0-255
rounded_data = round(scaled_data);  %rounds them down
data = dec2bin(rounded_data,8);     %convert the binary data to 8 bit binary #s

%open a file
output_name = strcat('./Images/',Number,'_rcm.coe');
file = fopen(output_name,'w');

%write the header info
fprintf(file,'memory_initialization_radix=2;\n');
fprintf(file,'memory_initialization_vector=\n');
fclose(file);

%put commas in the data
rowxcolumn = size(data);    
rows = rowxcolumn(1);
columns = rowxcolumn(2);
output = data;
for i = 1:(rows-1)
    output(i,(columns+1)) = ',';
end
output(rows,(columns+1)) = ';';

%append the numeric values to the file
dlmwrite(output_name,output,'-append','delimiter','', 'newline', 'pc');

% create color table for green (2)  and blue (3) and you're done!
green = color_table(:,2);             %grabs the red part of the colortable
scaled_data = green*255;              %scales the floats back to 0-255
rounded_data = round(scaled_data);  %rounds them down
data = dec2bin(rounded_data,8);     %convert the binary data to 8 bit binary #s

%open a file
output_name = strcat('./Images/',Number,'_gcm.coe');
file = fopen(output_name,'w');

%write the header info
fprintf(file,'memory_initialization_radix=2;\n');
fprintf(file,'memory_initialization_vector=\n');
fclose(file);

%put commas in the data
rowxcolumn = size(data);    
rows = rowxcolumn(1);
columns = rowxcolumn(2);
output = data;
for i = 1:(rows-1)
    output(i,(columns+1)) = ',';
end
output(rows,(columns+1)) = ';';

%append the numeric values to the file
dlmwrite(output_name,output,'-append','delimiter','', 'newline', 'pc');

%Blue
blue = color_table(:,3);             %grabs the red part of the colortable
scaled_data = blue*255;              %scales the floats back to 0-255
rounded_data = round(scaled_data);  %rounds them down
data = dec2bin(rounded_data,8);     %convert the binary data to 8 bit binary #s

%open a file
output_name = strcat('./Images/',Number,'_bcm.coe');
file = fopen(output_name,'w');

%write the header info
fprintf(file,'memory_initialization_radix=2;\n');
fprintf(file,'memory_initialization_vector=\n');
fclose(file);

%put commas in the data
rowxcolumn = size(data);    
rows = rowxcolumn(1);
columns = rowxcolumn(2);
output = data;
for i = 1:(rows-1)
    output(i,(columns+1)) = ',';
end
output(rows,(columns+1)) = ';';

%append the numeric values to the file
dlmwrite(output_name,output,'-append','delimiter','', 'newline', 'pc');
%% Turning a 2D image into a 1D memory array
%The code above is all well and good for the color table, since it's 1-D
%(well, at least you can break it into 3 1-D arrays).  But what about a 2D
%array?  We need to turn it into a 1-D array:

picture_size = size(picture);       %figure out how big the image is
num_rows = picture_size(1);
num_columns = picture_size(2);

pixel_columns = zeros(picture_size(1)*picture_size(2),1,'uint8');  %pre-allocate a space for a new column vector

for r = 1:num_rows
    for c = 1:num_columns
        pixel_columns((r-1)*num_columns+c) = picture(r,c);    %pixel# = (y*numColumns)+x
    end
end

%so now pixel_columns is a column vector of the pixel values in the image
rounded_data = round(pixel_columns);  %rounds them down
data = dec2bin(rounded_data,8);     %convert the binary data to 8 bit binary #s

%open a file
output_name = strcat('./Images/',Number,'.coe');
file = fopen(output_name,'w');

%write the header info
fprintf(file,'memory_initialization_radix=2;\n');
fprintf(file,'memory_initialization_vector=\n');
fclose(file);

%put commas in the data
rowxcolumn = size(data);
rows = rowxcolumn(1);
columns = rowxcolumn(2);
output = data;
for i = 1:(rows-1)
    output(i,(columns+1)) = ',';
end
output(rows,(columns+1)) = ';';

%append the numeric values to the file
dlmwrite(output_name,output,'-append','delimiter','', 'newline', 'pc');

%just to make sure that we're doing things correctly
regen_picture = zeros(num_rows,num_columns,'uint8');
for r = 1:num_rows
    for c = 1:num_columns
        regen_picture(r,c) = pixel_columns((r-1)*num_columns+c,1);
    end
end

figure
subplot(121)
image(picture)
axis square
colormap(color_table)
colorbar                            
title('Original Picture')

subplot(122)
image(regen_picture)
axis square
colormap(color_table)
colorbar                            
title('Regenerated Picture')