Threshold based segmentation will not yield good results if the features of interest cannot be easily distinguished using the histogram of pixel values. For example, grains in a microscope image (or cells) will not be efficiently separated thus resulting in wrong insights about the sample.

Watershed assisted segmentation is ideal for these situations. The image can be thresholded first using traditional approach to identify definitely positive and definitely negative regions. Then, watershed algorithm can be used to fill the gaps. This video demonstrates the use of watershed algorithm using a microscope image showing grains and boundaries (same example as previous video in this playlist).

The code from this video is available at:

Subscribe :


Grain Size Analysis in Python Using Watershed
4.35 GEEK