Generate coe file with 12 bits

I have a image on 24 bits and i want to have a coe file with 12 bits on a row.

This is the code:

clear all;
clc;
close all; 

[filename, pathname] = uigetfile('Capture.PNG','Pick an M-file');
imagine = imread(filename);
[ncolumns,nrows,ncolors]=size(imagine);

im_12bit (:,:,1) = dec2bin ( round (imagine (:,:,1)/16),4);
im_12bit (:,:,2) = dec2bin ( round (imagine (:,:,2)/16),4);
im_12bit (:,:,3) = dec2bin ( round (imagine (:,:,3)/16),4);
 

fid = fopen('out3333.coe', 'w'); %Create COE file
fprintf(fid, 'memory_initialization_radix=2;\n');
fprintf(fid, 'memory_initialization_vector=\n');
m = size(imagine); % Get the image size, m(1) is high, m(2) is wide


for j =1 : m(1)
    for i =1 : m(2)
                 % Write RGB data together
               
       fprintf(fid, '%s%s%s,\n', im_12bit(j,i,1), im_12bit (j,i,2),im_12bit(j,i,3));
       
    end
end
fseek(fid, -2, 1); % Overwrite the last comma with a semicolon
fprintf(fid, ';');

i have this error:

Index in position 2 exceeds array bounds (must not exceed 5).

Can you help me to solve this? It s another way to do it?

Thanks.

Answer 1

[Walter Roberson]

Suppose you have an entry which is uint8(255) . Then

uint8(255)/16

ans = uint8 16

Notice this is already rounded -- the integer data types are defined to operate like

A/B --> round(double(A)/double(B)) and then make it the integer data type

Now let's have a look:

dec2bin(round(uint8(255)/16), 4)

ans = '10000'

Notice you got out 5 bits even though you only expected 4. To get out 4, you need to floor() instead of round -- and that implies you have to convert the image to double yourself, so that the default behavior of round() does not interfere.

Now the next thing you need to look at is

dec2bin([1 2; 3 4; 5 6], 4)

ans = 6×4 char array
    '0001'
    '0011'
    '0101'
    '0010'
    '0100'
    '0110'

You passed in a 3 x 2 array, but you got out a 6 x 4 array, where 4 is the number of bits per entry. When you use dec2bin() you do not get out string objects. The result of dec2bin() on that array is not going to be

["0001", "0010"; "0011", "0100"; "0101", "0110"]

ans = 3×2 string array
    "0001"    "0010"
    "0011"    "0100"
    "0101"    "0110"

so you are not going to be able to index the result of dec2bin() at a row and column number and expect the result to be the binary for the corresponding image row and column. You can arrange it to be that way though:

im = imresize(imread('flamingos.jpg'),[10 10]);
image(im)

bits = reshape(string(dec2bin( floor(double(im)/16), 4 )), size(im));
bits(1:2,1:2,:)

ans = 2×2×3 string array
ans(:,:,1) = 

    "0110"    "0111"
    "0111"    "0100"


ans(:,:,2) = 

    "0110"    "0111"
    "0100"    "0100"


ans(:,:,3) = 

    "0100"    "0101"
    "0011"    "0011"