Cluster analysis by binary morphology
J.‐G. Postaire, R.D. Zhang, C. Lecocq-Botte · IEEE Transactions on Pattern Analysis and Machine Intelligence · 1993
An approach to unsupervised pattern classification that is based on the use of mathematical morphology operations is developed. The way a set of multidimensional observations can be represented as a mathematical discrete binary set is shown. Clusters are then detected as well separated subsets by means of binary morphological transformations.>