What is the most efficient way to obtain the centroid of each cluster of centroids in the matrix produced by 'kmeans' cluster analysis?

I am counting geese in large flocks in aerial photos using the 'detectSURFFeatures' function to identify individual geese. This gives me multiple feature centroids clustered at each goose image, which I analyze with 'kmeans'. I am currently sorting the resulting matrix of centroid values on cluster labels and using loops to calculate the mean centroid value for each cluster (see attached code).

Answer 1

[Image Analyst]

The most efficient way would be to get the second return argument of kmeans(). I mean, it gives you the cluster centers so why not accept them?

Answer 2

[Tony]

That's what I thought reading the documentation, but wasn't sure. Thank you again for help with application of cluster analysis.