Suppose we have an image, like so:


and I managed to create its MorphologicalGraph by using

p1 = Binarize[*the image*];
p2 = Thinning[Pruning[Thinning[DeleteSmallComponents[ColorNegate[DeleteSmallComponents[p1, 300]], 300]]]]

Which gave the result of


If we overlay the two, we can see how most of it lines up. Is there a way to extract the white parts in the image individually? I'm thinking of using MorphologicalGraph to extract the enclosed spaces, but any method / tip that can achieve extracting the white parts individually is perfect and greatly appreciated.

For example, each of those numbered is a part that is to be extracted. example


2 Answers 2


One approach is to use:

MorphologicalComponents[Erosion[img, 1]] // Colorize

enter image description here

Then you can access the individual colored sections using ComponentMeasurements. For example:

ComponentMeasurements[comp // Colorize, {"Image", "Count", "Mean"}, All, "Dataset"]

gives a long list of all the segments and how large they are. You can sort them by many different properties. Here is a piece of the output:

enter image description here


You're looking for MorphologicalComponents[*the image*]. This function groups connected areas of white pixels and assigns them a single integer. For example;

mc = MorphologicalComponents[Binarize[*the image*]]


  Table[Style[Text[mc[[i, j]], {i, j}], 8], {i, 300, 350}, {j, 300, 
    350}], ImageSize -> 8 72]

gives the following subset of the mc data

enter image description here

As you can see, each cluster of white pixels now has a unique number, such as 81 in the case of the upper left corner of this graphic. The 0s correspond to the black areas. You can use SelectComponents[mc, -criteria-] to search for large, small, round, etc features. Below I choose the 10 largest.

Colorize@SelectComponents[mc, "Area", -10]

enter image description here

To see which integers are assigned to which cluster of white pixels, use the following;

centroids = ComponentMeasurements[mc, "Centroid"];

where centroids[[1]] gives 1 -> {674.491, 672.}

  Graphics[ {White,
   Table[ Text[centroids[[i, 1]], centroids[[i, 2]]], {i, Length@centroids}]}]

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.