Making measurements on images

This lesson was adapted from the Matlab Image segmentation tutorial which can be found here: http://www.mathworks.com/matlabcentral/fileexchange/25157-image-segmentation-tutorial?s_eid=PSM_gen

The challenge files for this part of the course can be found here. The images were obtained from here- http://spc.ucsd.edu/#!

Display the histogram of the image.

coins= imread('coins.png');% load in coins.png

We are going to do this by making a histogram. We will use the imhist function that comes with the image processing toolbox. Note: You can also use the inbuilt hist function.

[pixelcount, grayLevels]= imhist(coins); % pixelcount contains the intensity counts grayLevels contain the xaxis
figure
bar(pixelcount)

% Manually setting the x and the y limits
xlim([0 max(grayLevels)]); 
ylim([0 max(pixelcount(:))]);

Challenge 1

Display the intensity of pixel histogram for the image file - plankton

% Read in the plankton image
% Get the histogram data and the grayscale data
% display the histogram

Converting image to binary

Look through the histogram to set a threshold. It looks like there are two clear peaks, one for all the darker values and one for the larger values. We can convert an image to binary using many methods. The simplest way of doing this is by using conditionals.

Threshold= 100; % Set a threshold.
binary_coins= coins> Threshold; % In the variable binary_coins, all the
% values in coins greater than the threshold will be made 1. Everything
% else will be made 0.

We can invert the image using ~. This will make the ones=zero and zeros=ones.

binary_coins_inverted= ~binary_coins; % invert the array 

subplot(1,2,1) % plotting the two images side by side.
imshow(binary_coins)

subplot(1,2,2)
imshow(binary_coins_inverted)

Challenge 2

Set a threshold and convert the plankton into binary image.

% Define the threshold for the plankton image
% Use conditional formatting to convert to binary

Filling in the gaps in our image.

Since our object is to identify objects in our image, we have to make sure that the objects are uniform.

fill_coins= imfill(binary_coins, 'holes'); % If we leave out 'holes', then imfill opens
% an interactive interface where we can specify which gaps to fill.
figure; imshow(fill_coins);

We can see that the gaps have now been filled in.

Challenge 3

Use imfill to fill in the gaps in the binary image of the plankton

Finding and labelling connected objects

We can use the bwlabel function to give each of the objects in an image a a unique identitiy.

[labeled_coins, count]=bwlabel(fill_coins);%label the different elements rgb_labeledcoins=label2rgb(labeled_coins);%convert labeled image to rgb imagesc(rgb_labeledcoins);%display labeled image impixelinfo% get pixel information.

imagesc(labeled_coins); colormap gray % display the same image in gray scale impixelinfo %get pixel information.

If we hover over each of the coins, they will have a number corresponding to their label. We can isolate each of the objects once again using our best friends: conditionals. We can display only one of the coins.

coin_1= labeled_coins==1;%select just the coin labelled 1.
imagesc(coin_1); colormap gray; %display the single coin

We can get all the individual coins as separate element of a structural array


connected_coins= bwconncomp(fill_coins, 8) % Don't suppress output
connected_coins.PixelIdxList;

We can see that each of the coins are a separate cell in the structural array connected_coins.PixelIdxList.

Challenge 4

Find all the connected objects in the filled plankton image.