How to divide an image into non-overlapping blocks size 4x4?

I want to divide a color image into a series of non-overlapping blocks size 4x4. Then, i want to know the minimum pixel value each block and arrange it into matrix just like before.

Can anyone help, please.

[Jan]

What is "the minimum pixel value" for an RGB pixel, which consists of 3 components?

[Anisa]

Sorry, actually i'm working on grayscale image but my lecturer want me to do that on RGB image. So here i am, trying to find every possible answer.

[Jan]

It is impossible to post "every possible answer". There is an an enormously large number of possibilities to define a norm, which is required to determine a "minimum". It would be much more efficient when you find out by your own, what you or your professor wants. So please explain: What is the minimum pixel value between two [1x3] RGB colors?

[Anisa]

I'd edited my question. I want to divide a grayscale image of size 512x512 named Lena.bmp into a series of non-overlapping blocks of size 16x16 blocks. Then, i want to know the minimum pixel value each block and save it into matrix 16x16. I have try but it didn't work. When i checked the result between command window and workspace, it has different number.

   I=imread('Lena.bmp', 'bmp');
   c=mat2cell(I, [32 32 32 32 32 32 32 32 32 32 32 32 32 32 32 32], ...
              [32 32 32 32 32 32 32 32 32 32 32 32 32 32 32 32]);
   minTemp=zeros(16,16);
for i=1:16;
    for j=1:16;

        a = c{i,j};
        min=a(1,1);

        for s=1:32;
            for t=1:32;
                if a(s,t)<min
                   min = a(s,t);
                end
            end
        end

        minTemp(i,j)=min

    end
end

Help me, please.

[Jan]

Using the command "min" would be much more efficient than using this as name of a variable in a double FOR loop:

minTemp(i,j) = min(c{i,j}(:));

But see the [EDITED] section in my answer for a faster and nicer solution.

Answer 1

[Image Analyst]

Anisa: Regarding your latest edit. If you want to do it that way, then you don't need the innermost two loops. Simply do it like this:

I=imread('lena.bmp');
c = mat2cell(I, [32 32 32 32 32 32 32 32 32 32 32 32 32 32 32 32], ...
  [32 32 32 32 32 32 32 32 32 32 32 32 32 32 32 32]);
minOfBlocks=zeros(16,16);
subplot(1,2,1);
imshow(I);
title('Original Image', 'FontSize', 30);
for i=1:16
  for j=1:16
    a = c{i,j};
    minOfBlocks(i,j)=min(a(:));
  end
end
subplot(1,2, 2);
imshow(minOfBlocks, []);
title('Min of 32x32 Blocks', 'FontSize', 30);

[Anisa]

Yup, the results same with mine. But, when i check the minimum value between minOfBlocks and variable editor of c, it's different.

For example, it's 148 on minOfBlocks(1,1). Then, when i check on 32x32 of c{1,1} there's no 148. How do i know that minimum value on minOfBlocks(1,1) same with the 'real' minimum value on c{1,1}?

Thanks you.

[Image Analyst]

It should be the same. Please add appropriate fprintf()'s to prove otherwise. Also you can upload your image to tinypic.com (or wherever) if you want me to use your image, or suggest a standard MATLAB demo image, like cameraman or moon or something (may need a resize to get to 512x512 though).

[RAKESH KUCHANA]

Hello Image Analyst, I want to split an image of 178*665 pixel format into 3 parts. Can you provide a code for this?

Answer 2

[Jan]

No loops, no blockproc:

RGB = rand(200, 100, 3);  % Example data
s   = size(RGB);
Blocks = reshape(RGB, [4, s(1)/4, 4, s(2)/4, 3]);
Blocks = min(min(Blocks, 1), 3);

But what does "arrange it into a matrix just like before"? Perhaps:

Blocks = repmat(Blocks, [4, 1, 4, 1, 1]);
Result = reshape(Blocks, s);

Now the non-overlapping 4x4 blocks have the same color, which is the minimal R, G and B component of the corresponding pixels.

[EDITED after Anisa's comment]

Im       = imread('Lena.bmp');
blockIm  = reshape(Im, 16, 32, 16, 32);
blockMin = min(min(blockIm, [], 1), [], 3);
Result   = reshape(blockMin, 32, 32);

[Jan]

What's wrong with this easier, faster and loop-free method?

[sumit kumar]

i have a image and want to divide in 16*16 block and then want all the blocks separately.How will i proceed.??

[Image Analyst]

sumit: See the FAQ: http://matlab.wikia.com/wiki/FAQ#How_do_I_split_an_image_into_non-overlapping_blocks.3F

Answer 3

[David]

How does this work for you?

I = imread('AnisasImage.jpg');
I = rgb2gray(I);

for col = 1:4:size(I,2)-4
  for row = 1:4:size(I,1)-4

    block_minimums((row+3)/4,(col+3)/4) = min(min((I(row:row+4,col:col+4))));

    end
  end

[David]

You wanted colour didn't you.

I = imread('AnisasImage.jpg');
for col = 1:4:size(I,2)-4
  for row = 1:4:size(I,1)-4
    block_minimums((row+3)/4,(col+3)/4) = min(min(min((I(row:row+4,col:col+4,:)))));
    end
  end

Answer 4

[Image Analyst]

Anisa, try this demo (that I've posted many times before). Just copy, paste, and run. It is pretty general and flexible. It splits both color and grayscale standard MATLAB demo images up into blocks, using blockproc() and using regular indexing (both methods). It even allows you to have images that are not a integer multiple of the block size and that is illustrated. So it pretty much handles every 2D situation.

% Demo to divide a color image up into blocks.
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
workspace;  % Make sure the workspace panel is showing.
fontSize = 20;

% Read in a standard MATLAB color demo image.
folder = fullfile(matlabroot, '\toolbox\images\imdemos');
baseFileName = 'peppers.png';
% Get the full filename, with path prepended.
fullFileName = fullfile(folder, baseFileName);
if ~exist(fullFileName, 'file')
  % Didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
% Read the image from disk.
rgbImage = imread(fullFileName);

% Test code if you want to try it with a gray scale image.
% Uncomment line below if you want to see how it works with a gray scale image.
% rgbImage = rgb2gray(rgbImage);

% Display image full screen.
imshow(rgbImage);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);
drawnow;
% Get the dimensions of the image.  numberOfColorBands should be = 3.
[rows columns numberOfColorBands] = size(rgbImage)

%==========================================================================
% The first way to divide an image up into blocks is by using mat2cell().
blockSizeR = 150; % Rows in block.
blockSizeC = 100; % Columns in block.

% Figure out the size of each block in rows. 
% Most will be blockSizeR but there may be a remainder amount of less than that.
wholeBlockRows = floor(rows / blockSizeR);
blockVectorR = [blockSizeR * ones(1, wholeBlockRows), rem(rows, blockSizeR)];
% Figure out the size of each block in columns. 
wholeBlockCols = floor(columns / blockSizeC);
blockVectorC = [blockSizeC * ones(1, wholeBlockCols), rem(columns, blockSizeC)];

% Create the cell array, ca.  
% Each cell (except for the remainder cells at the end of the image)
% in the array contains a blockSizeR by blockSizeC by 3 color array.
% This line is where the image is actually divided up into blocks.
if numberOfColorBands > 1
  % It's a color image.
  ca = mat2cell(rgbImage, blockVectorR, blockVectorC, numberOfColorBands);
else
  ca = mat2cell(rgbImage, blockVectorR, blockVectorC);
end

% Now display all the blocks.
plotIndex = 1;
numPlotsR = size(ca, 1);
numPlotsC = size(ca, 2);
for r = 1 : numPlotsR
  for c = 1 : numPlotsC
    fprintf('plotindex = %d,   c=%d, r=%d\n', plotIndex, c, r);
    % Specify the location for display of the image.
    subplot(numPlotsR, numPlotsC, plotIndex);
    % Extract the numerical array out of the cell
    % just for tutorial purposes.
    rgbBlock = ca{r,c};
    imshow(rgbBlock); % Could call imshow(ca{r,c}) if you wanted to.
    [rowsB columnsB numberOfColorBandsB] = size(rgbBlock);
    % Make the caption the block number.
    caption = sprintf('Block #%d of %d\n%d rows by %d columns', ...
      plotIndex, numPlotsR*numPlotsC, rowsB, columnsB);
    title(caption);
    drawnow;
    % Increment the subplot to the next location.
    plotIndex = plotIndex + 1;
  end
end

% Display the original image in the upper left.
subplot(4, 6, 1);
imshow(rgbImage);
title('Original Image');

%==============================================================================
% Another way to split the image up into blocks is to use indexing.
% Read in a standard MATLAB gray scale demo image.
folder = fullfile(matlabroot, '\toolbox\images\imdemos');
baseFileName = 'cameraman.tif';
fullFileName = fullfile(folder, baseFileName);
% Get the full filename, with path prepended.
fullFileName = fullfile(folder, baseFileName);
if ~exist(fullFileName, 'file')
  % Didn't find it there.  Check the search path for it.
  fullFileName = baseFileName; % No path this time.
  if ~exist(fullFileName, 'file')
    % Still didn't find it.  Alert user.
    errorMessage = sprintf('Error: %s does not exist.', fullFileName);
    uiwait(warndlg(errorMessage));
    return;
  end
end
grayImage = imread(fullFileName);
% Get the dimensions of the image.  numberOfColorBands should be = 1.
[rows columns numberOfColorBands] = size(grayImage);
% Display the original gray scale image.
figure;
subplot(2, 2, 1);
imshow(grayImage, []);
title('Original Grayscale Image', 'FontSize', fontSize);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);

% Divide the image up into 4 blocks.
% Let's assume we know the block size and that all blocks will be the same size.
blockSizeR = 128; % Rows in block.
blockSizeC = 128; % Columns in block.
% Figure out the size of each block. 
wholeBlockRows = floor(rows / blockSizeR);
wholeBlockCols = floor(columns / blockSizeC);
% Preallocate a 3D image
image3d = zeros(wholeBlockRows, wholeBlockCols, 3);
% Now scan though, getting each block and putting it as a slice of a 3D array.
sliceNumber = 1;
for row = 1 : blockSizeR : rows
  for col = 1 : blockSizeC : columns
    % Let's be a little explicit here in our variables
    % to make it easier to see what's going on.
    row1 = row;
    row2 = row1 + blockSizeR - 1;
    col1 = col;
    col2 = col1 + blockSizeC - 1;
    % Extract out the block into a single subimage.
    oneBlock = grayImage(row1:row2, col1:col2);
    % Specify the location for display of the image.
    subplot(2, 2, sliceNumber);
    imshow(oneBlock);
    % Make the caption the block number.
    caption = sprintf('Block #%d of 4', sliceNumber);
    title(caption);
    drawnow;
    % Assign this slice to the image we just extracted.
    image3D(:, :, sliceNumber) = oneBlock;
    sliceNumber = sliceNumber + 1;
  end
end
% Now image3D is a 3D image where each slice, 
% or plane, is one quadrant of the original 2D image.

msgbox('Done with demo!  Check out the two figures.');

[Im@]

hello I would like you please to provide where i can change in the code the number of blocks i want

[Image Analyst]

Change these lines:

blockSizeR = 150; % Rows in block.
blockSizeC = 100; % Columns in block.

Answer 5

[Matt J]

Here's a way using this FEX file,

http://www.mathworks.com/matlabcentral/fileexchange/35085-split-nd-array-into-equal-sized-cells

though I don't know if you're permitted to use it for course-work

out=cellfun(@(c) min(c(:)), mat2tiles(yourImage,[16,16]));

Answer 6

[Jos (10584)]

What about using the dedicated tool BLOCKPROC (provided you have the Image Processing Toolbox) which gives the following one-liner:

OUT = blockproc(IN, [4 4],  @(BS) repmat(min(BS.data(:)), size(BS.data)))

Answer 7

[Matt J]

You can use sepblockfun

https://www.mathworks.com/matlabcentral/fileexchange/48089-separable-block-wise-operations

out =sepblockfun(yourImage,[4,4,inf],'min');