Image Features - First Order Statistics - Weighted Centroid


I’m using the node Image Features to extract features (First Order statistics).

What are the weighted centroid Dim 1-2-3-4-5 ?
How are they calculated ?

Thanks for the support

Hi @trj,

The weighted centroid computation is based on a “regular” centroid computation: Instead of computing the arithmetic mean of all the point coordinates of an object, the point coordinates are weighted by their intensity. That is, regions with high intensities/values that are closer to the bounds will pull the weighted centroid away from the centroid of the image (which is its center for rectangular images). The Dim 1, Dim 2, etc., are the individual values per dimension of the image: Dim 1 and Dim 2 are x and y, respectively. Other dimensions (e.g. channels, time, slices, …), will be reported as 0 if they are not available in your input image.

That’s the implementation of the computation:



This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.