Watershed Transforms

Watershed transforms can be a powerful technique for separating the overlapping objects in an image. This technique uses the labels of a multi-ROI as the seed points and a landscape image whose 'catchment basins' correspond to the objects you want to identify. In the example below, a watershed transform was applied to segment the grains in a sandpack dataset.

Original image (left) and the result of applying a watershed transform to segment the grains (right)

Watershed transform result

This video introduces users to the Watershed Transform function in Dragonfly.

Watershed Transform video tutorial (39:36)

You can also view this video and others on our YouTube channel (www.youtube.com/channel/UCuFl2zHcyStR2RJpMXbi6ow).

You should first complete all of the required pre-processing steps before you attempt to apply a watershed transform to your data. This can include creating a landscape image whose 'catchment basins' are the objects that you want to identify, creating a multi-ROI with the required seed points, and creating a mask to define the area of application for the watershed transform. Refer to the topics in Pre-Processing Steps for information about these steps.

How to Apply a Watershed Transform
  1. Right-click the multi-ROI that will provide the seed points and then choose Watershed Transform in the pop-up menu.

    Note Refer to the topic Seeds for information about creating multi-ROIs that will provide the seed points.

  2. Choose the dataset that will provide the local topography for the watershed transform in the Choose a landscape dialog, as shown below.

    Note Refer to the topic Landscape Images for information about creating landscape images.

  3. Click OK.

    The Choose an ROI as Mask dialog appears.

  4. Choose a mask in the Object drop-down menu, optional.

    Note Masks let you define the area of application for the watershed transform. Refer to the topic Creating Masks for information about creating masks.

  5. Click OK.

    The watershed is evaluated with the seeds and the selected landscape image. Wait for the expanded objects to be computed and then overwritten into the initial multi-ROI.