How is distance to neighbour objects measured?
I'm fairly new to eCognition but wondering if you can help.
I'm classifying a Sentinel 2 image and trying to add a rule of 'distance to class', assuming that it would measure the nearest point of an object to the border of a class. In this case I was trying to classify 'shoreline' by being no more than 2 pixels from a classification 'water'. However the distances don't make sense unless it's measuring to a class centroid or something. Objects that share a border with water are shown as a distance over 80m or 8 pixels but further objects are sometimes less.
There are no layers with different projections in the project.
Can anyone help me?
Was this article helpful?
Please try to change the Distance Calculation to Smallest Enclosing Rectangle in the algorithm set rule set options.
More information about the available methods can be found in the User Guide > Introduction > Image Object related Features.