Unsupervised image segmentation based on the comparison of local and regional histograms
A.A. Dingle, M. Morrison · 2002
This paper proposes an new method for unsupervised segmentation of images which does not rely on parametric modelling of the observed images. Furthermore, the problem of finding the number of image classes is carried out as an integral part of the segmentation process, rather than by resorting to goodness-of-fit cluster validation measures, such as AIC or MDL. A brief overview of the algorithm is given, as well as examples of its application to both synthetic and real images.