Segmenting hyperspectral images is an important task for simplifying the analysis of the data by focusing on a certain part of the data set or on data samples of the same or at least "nearby" spectral properties. A new class of neuro-fuzzy systems, based on so-called weighted incremental neural networks (WINN), is briefly introduced, exemplified and finally used for unsupervised segmentation of hyperspectral images. The WINN algorithm produces a net of nodes connected by edges, which reflects and preserves the topology of the input data set. Additional weights, which are proportional to the local data densities in the input space, are associated with the resulting nodes and edges to store useful information about the topological relations in the given input data set. A fuzziness factor, related to the connectedness of the net, is introduced in the system. A watershed-like procedure is used to cluster the resulting net. The number of the resulting clusters is determined by this procedure. Experimental results underline the usefulness and efficiency of the proposed neuro-fuzzy system for multi-dimensional data clustering and image segmentation, especially hyperspectral images.
Unsupervised hyperspectral image segmentation using a new class of neuro-fuzzy systems based on weighted incremental neural networks
01.01.2002
4371331 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 2002
|Unsupervised Fuzzy Clustering and Image Segmentation Using Weighted Neural Networks
British Library Conference Proceedings | 2003
|A latent class model with fuzzy segmentation and weighted variables
Taylor & Francis Verlag | 2014
|Unsupervised multiscale image segmentation
IEEE | 1999
|