
This example shows how to expand seeds dependent on the individual seed characteristics (local conditions) using pixel-based object resizing instead of introducing an object neighborhood and using "image object fusion"-based methods.
Spectral object pixel features, object shape and the whole bandwith of Definiens Developer object features can be used to restrict or control the reshaping process. The algorithms "Pixel-based Object Resizing" and "Multi Threshold Segmentation" support object variables as local condition.
The method could look like this in a word model:
1) find seeds
2) resize seeds due to the individual seed object conditions:
Example 1: shrink seeds and remove parts that are brighter than twice the minimum value in layer 1 (two Alternatives using "Pixel-based Object Resizing" and "Multi Threshold Segmentation")
Example 2: Shrink seeds to 25% of their initial area
Please refer to the ruleset process comments for more detailed explanations of the CNL implementation.
Comments