Blob segmentation and writing it to folder

i have segmented few blobs using the below code

ccnt = 1;
for blob = 1 : n
    thisBoundingBox = measurements(blob).BoundingBox;
    [new_blob, pos] = imcrop(binaryImage, thisBoundingBox);
    fileName = sprintf('FolderName/%d.bmp',ccnt);
    imwrite(new_blob, fileName); ccnt = ccnt + 1;
end

When using the above code, i get all the blobs, but the order of the blobs written in the folder is not as i want

If my image is

i want each segmented blob to be written in the folder in the left to right order as

but i get it as

what should i do to get it in left to right order

Answer 1

[Image Analyst]

m = [...
	1,2,3,4;
	5,6,7,8;
	9,10,11,12]
m2 = reshape(m', 1, []) % Transpose then reshape into row vector.

m =

1 2 3 4

5 6 7 8

9 10 11 12

m2 =

1 2 3 4 5 6 7 8 9 10 11 12

[Elysi Cochin]

sir m is not a matrix, its an image.

[Image Analyst]

It does not matter. I'm sure you know that an image IS a matrix, and vice versa : a matrix can be considered as an image.

[Elysi Cochin]

sir, i have edited the question, please check if my input is such an image, what can be done to write the segmented blobs in left to right order, line by line

[Image Analyst]

I'd use regionprops to get the centroids. Then knowing there are 2 blobs in each row, you can figure out which blob is in which quadrant.

props = regionprops(mask, 'Centroid', 'BoundingBox');
xy = vertcat(props.Centroid);
x = xy(:, 1);
meanx = mean(x)
y = xy(:, 2);
meany = mean(y);
for k = 1 : length(props)
    thisx = x(k)
    thisy = y(k)
    if thisx < meanx && thisy < meany
        % It's in quadrant 1
        fileName = '1.png'
    elseif thisx > meanx && thisy < meany
        % It's in quadrant 2
        fileName = '2.png'
    elseif thisx < meanx && thisy > meany
        % It's in quadrant 3
        % etc.
    end
    croppedImage = imcrop(mask, props(k).BoundingBox)
    imwrite(croppedImage, fileName);
end

See if you can finish it.

[Elysi Cochin]

what if there are more than 2 blobs ( more number of rows and columns ). The number of blobs in the image keeps changing.

Sir, also that code didnt work for me. This is the order i got, but the first image should be STAR

[Image Analyst]

Elysi then you didn't complete it correctly. Perhaps you were thrown off by the fact that there are there are two dots in the star. I handled that by filling the blobs. If you don't want it filled, then you can keep them using a few other special operations. Here is the way that you should have completed the code:

clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 22;

%--------------------------------------------------------------------------------------------------------
%    READ IN IMAGE
folder = pwd;
baseFileName = 'image.png';
% Get the full filename, with path prepended.
fullFileName = fullfile(folder, baseFileName);
% Check if file exists.
if ~exist(fullFileName, 'file')
	% The file doesn't exist -- didn't find it there in that folder.
	% Check the entire search path (other folders) for the file by stripping off the folder.
	fullFileNameOnSearchPath = baseFileName; % No path this time.
	if ~exist(fullFileNameOnSearchPath, 'file')
		% Still didn't find it.  Alert user.
		errorMessage = sprintf('Error: %s does not exist in the search path folders.', fullFileName);
		uiwait(warndlg(errorMessage));
		return;
	end
end
grayImage = imread(fullFileName);

% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(grayImage);
if numberOfColorChannels > 1
	% It's not really gray scale like we expected - it's color.
	% Use weighted sum of ALL channels to create a gray scale image.
	grayImage = rgb2gray(grayImage);
	% ALTERNATE METHOD: Convert it to gray scale by taking only the green channel,
	% which in a typical snapshot will be the least noisy channel.
	% grayImage = grayImage(:, :, 2); % Take green channel.
end
% Display the image.
subplot(2, 2, 1);
imshow(grayImage, []);
title('Original Grayscale Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
hFig = gcf;
hFig.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.
drawnow;


%--------------------------------------------------------------------------------------------------------
%    SEGMENTATION OF IMAGE
% Get a binary image
binaryImage = ~imbinarize(grayImage);
% Get rid of the small dots in the star by doing a hole fill.
binaryImage = imfill(binaryImage, 'holes');
subplot(2, 2, 2);
% figure
imshow(binaryImage, []);
impixelinfo;
title('Binary Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;

%--------------------------------------------------------------------------------------------------------
% MEASUREMENT OF CENTROIDS
% Measurement of locations and assignment of file names.
props = regionprops(binaryImage, 'Centroid', 'BoundingBox');
xy = vertcat(props.Centroid);
x = xy(:, 1);
meanx = mean(x)
y = xy(:, 2);
meany = mean(y)

%--------------------------------------------------------------------------------------------------------
% PLOTTING AND SAVING
% Put up lines there.
yline(meany, 'LineWidth', 2, 'Color', 'r');
xline(meanx, 'LineWidth', 2, 'Color', 'r');
for k = 1 : length(props)
	thisx = x(k);
	thisy = y(k);
	% Plot a red star at the centroid.
	hold on;
	plot(thisx, thisy, 'r*', 'MarkerSize', 15, 'LineWidth', 2);
	if thisx < meanx && thisy < meany
		% It's in quadrant 1
		fileName = '1.png';
	elseif thisx > meanx && thisy < meany
		% It's in quadrant 2
		fileName = '2.png';
	elseif thisx < meanx && thisy > meany
		% It's in quadrant 3
		fileName = '3.png';
	elseif thisx > meanx && thisy > meany
		% It's in quadrant 4
		fileName = '4.png';
	end
	croppedImage = imcrop(binaryImage, props(k).BoundingBox);
	% Save the image.
	fprintf('Blob #%d at (%.1f, %.1f) is being saved as %s.\n', k, thisx, thisy, fileName);
 	imwrite(croppedImage, fileName);
	% Display images and labels.
	subplot(2, 4, k+4);
	imshow(croppedImage);
	title(fileName, 'Color', 'r', 'FontSize', fontSize);
	% Put a label over the blob in the upper right image.
	subplot(2, 2, 2);
	label = sprintf('Blob %d = %s', k, fileName);
	text(thisx, thisy + 8, label, 'Color', 'r', 'FontSize', 13, 'FontWeight', 'bold', 'HorizontalAlignment', 'center');
end
impixelinfo;

You can see that the blobs have the correct filenames.

[Elysi Cochin]

Sir the previous example (2 rows 2 columns) is working. But what if i have more number of rows or columns (number of rows and columns differ in images), like

where in the loop should i make the change to get the output as

[Image Analyst]

Then you need to use kmeans on the x and y centroids to find out how many rows the blobs are in. Then get inbetween the centroids as the dividing lines.

[indexes, rowCentroids] = kmeans(yCentroid, 3);

If you don't know how many rows there are, you can also tell kmeans to figure it out for you.

For the general case of blobs scattered around like a shotgun blast, then I don't see any reason or need to sort them - it would just be arbitrary and the default ordering (top-down, left-to-right) is as good as any ordering.

[Elysi Cochin]

sir how can kmeans find the number of rows?

In case of blobs scattered around like a shotgun blast, it doesnt matter, but i'm doing it in alphabets, then i need to identify the words, thats why

[Image Analyst]

In that case, lines of text like scanned in from a page of a book, I'd use my attached demo. Basically get a vertical profile and threshold to find the number of lines.

[Image Analyst]

I think you'd be best off picking a handwriting algorithm here: VisionBib

There are papers there that have successfully done it.

[Elysi Cochin]

Sir i found a paper, using histogram peak