Histogram Clustering Validation for Multispectral Image
V. S. Sidorova
Institute of Computational Mathematics and Mathematical Geophysics, Siberian Branch of the Russian Academy of Sciences, Novosibirsk, Russia, E-mail: svs@ooi.sscc.ru
Pages: 28-32
Abstract
A measure of assessment of uncontrolled space image classification by a clustering technique based on a multivariate histogram is proposed. The measure is used to choose the best vector partitions corresponding to different detail of data representation. It is experimentally shown that well-separated unimodal clusters of the partitions correspond to representative information classes of a sensed land cover.
|
