Unsupervised classification for multi-sensor data in remote sensing using Markov random field and maximum entropy method

Employs a multi-stage algorithm that makes use of spatial contextual information in a hierarchical clustering procedure for unsupervised image segmentation. The hierarchical clustering algorithm is based on similarity measures between all pairs of candidates being considered for merging. The multi-stage algorithm involves a local segmentor and a global segmentor. The data from individual sensors are integrated into a set of multidimensional data and it is then applied to the hierarchical clustering algorithm based on linear statistics under the assumption of an additive noise model.