Hierarchical classification of land-cover types using RAG-based merging

A multistage hierarchical clustering technique, which is an unsupervised technique, was suggested in this paper for classifying large remotely-sensed imagery. The multistage algorithm consists of two stages. The "local" segmentor of the first stage performs region-growing segmentation by employing a RAG-based merging with the restriction that pixels in a cluster must be spatially contiguous. The "global" segmentor of the second stage, which has not spatial constraints for merging, clusters the segments resulting from the previous stage. The second stage is an agglomerative hierarchical clustering procedure which merges the best MCN defined in spectral space, and then generates a dendrogram which represents a hierarchy of consecutive merging processes. The experimental results show that the new approach proposed in this study is quite efficient to analyze very large images. The technique was then applied to classify the land-cover types using the remotely-sensed data acquired from the Korean peninsula.