Please I want to modify that code for run length to work on gray scale image

   ------------------------------------------------------------------------

I want to modify that code to make it work on grayscale image not binary image can anyone help me

----------------------------------------------------------------------------------

function out=rle(image)
%
% RLE(IMAGE) produces a vector containing the run-length encoding of
% IMAGE, which should be a binary image. The image is set out as a long
% row, and the conde contains the number of zeros, followed by the number
% of ones, alternating.
%
% Example:
%
% rle([1 1 1 0 0;0 0 1 1 1;1 1 0 0 0])
%
% ans =
%
% 03453
%
level = graythresh(image);
BW = im2bw(image, level);
L=prod(size(BW));
im=reshape(BW',1,L);
x=1;
out=[];
while L ~= 0,
temp=min(find(im == x));
if isempty(temp),
out=[out L];
break
end;
out=[out temp-1];
x=1-x;
im=im(temp:L);
L=L-temp+1;
end;

------------------------------------------------------------------------

Please I want to modify that code to make it work on grayscale image not binary image can anyone help me

[Walter Roberson]

https://www.mathworks.com/matlabcentral/fileexchange/41813-runlength

[haithem abdelghany]

This code work on vector only i want to run the code on grayscale image

Answer 1

[Jan]

There is no Run Length defined for matrices, but only for vectors. Even your code converts the input image to a vector at first in:

im=reshape(BW',1,L);

So simply call FEX: RunLength by:

img    = rand(640, 480);         % Example image data
imgV   = reshape(img.', 1, []);  % Convert to vector
[B, N] = RunLength(imgV);

If you do not have a compiler installed or speed doesn't matter, user the included RunLength_M. Or cut the needed lines only:

function [b, n] = rle(ImageMatrix)
x = reshape(ImageMatrix.', [], 1);
d = [true; diff(x) ~= 0];   % TRUE if values change
b = x(d);                   % Elements without repetitions
k = find([d', true]);       % Indices of changes
n = diff(k);                % Number of repetitions
end

By the way: Your code is not a Run Length Encoder. The example from the help section

rle([1 1 1 0 0;0 0 1 1 1;1 1 0 0 0])

does not reply 03453, but only 3. This means that the posted code does not work. In consequence I cannot replace it directly by a method working with gray scale images. The commands graythresh and im2bw seems, like the code should work with grey scale images already.

[haithem abdelghany]

Thank u that is fine It works But how to calculate the compression ratio of the image after encoding with run-length with this code I mean using B or N

[Walter Roberson]

I discuss computation of compression ratio at https://www.mathworks.com/matlabcentral/answers/27322-cr-bit-per-pixel-image-compression#comment_410392

[haithem abdelghany]

Thank you I already use imratio function from raul gonzalez book Digital Image Processing using matlab I want to know which of output to use from the function to calculate the compression ratio B or N

[Jan]

You need both. The original image consists of numel(BW) pixels (find out if they are treated as bits or bytes), the compressed image uses the two arrays B and N. Now B has the same type as the input, but N is a double array, which requires 8 bytes per value. for a real image compression, it is more useful to limit N to 255, such that it can be stored in a UINT8:

[B, N] = RunLength(imgV, 'byte')

Now the compression factor is:

(numel(B) + numel(N)) / numel(imgV)

[haithem abdelghany]

It is ok but the compression ratio is greater than one which mean the size of the original file increase ?

[Jan]

Yes, this is possible. If each pixel is repeated twice, the compression is e.g. 1.0. For an image without repetitions, the output is twice as large. Therefore images are not compressed by pure RLE methods, as you know.

[haithem abdelghany]

Thank you I understand