Hubert's Gamma statistic applied here should be 0 for the case of either 1 cluster for everything, or every pixel in a separate cluster.
However, the search pattern used to find the best clustering parameter may not be returning non-zero Gamma statistics, leading to early exit with a terrible clustering solution.
Need to investigate and find solutions.
Low priority though, because it works more often than not.
Hubert's Gamma statistic applied here should be 0 for the case of either 1 cluster for everything, or every pixel in a separate cluster.
However, the search pattern used to find the best clustering parameter may not be returning non-zero Gamma statistics, leading to early exit with a terrible clustering solution.
Need to investigate and find solutions.
Low priority though, because it works more often than not.