Reducing the computation time of the Isodata and K-means unsupervised classification algorithms

Describes a sequence of methods to improve the running-time of the Isodata and K-means algorithms. The methods are compared on a range of remotely-sensed data sets, and show consistent and dramatic speedups.