Watershed Region Growth
Hi all,
I have recently been looking to do some watershed region growth within eCognition but have not found a way to do this yet.
Essentially, I am looking to use existing tree crowns, and grow/shrink these with an updated CHM. It is important that I retain the crown IDs, hence why I need to grow from existing polygons.
The pixel-based resizing algorithm is not quite what I am looking for as it takes in a hard threshold and could be missing out on local maxima/minima.
Has anyone ever done this?
Thanks!
Was this article helpful?
2 comments
Hello Loreena,
First of all, please check the 'watershed segmentation' algorithm, I believe it would be perfect for your task.
Here an introduction: https://support.ecognition.com/hc/en-us/articles/360016809079-Watershed-Segmentation
Here you can find two use cases:
If you want to generate your own segmentation approach, please spend some time in learning the 'image object fusion' algorithm. With the algorithm (in combination with a chessboard 1) you can define your own region growing approach.
But please try out the watershed segmentation first.
Regards,
Christian
Thank you Christian, I will have a look at these.