Classi ® cation by progressive generalization : a new automated methodology for remote sensing multichannel data

A new procedure for digital image classi® cation is described. The procedure, labelled Classi ® cation by Progressive Generalization (CPG), was developed to avoid drawbacks associated with most supervised and unsupervised classi® cations. Using lessons from visual image interpretation and map making, non-recursive CPG aims to identify all signi® cant spectral clusters within the scene to be classi® ed. The basic principles are: (i ) initial data compression using spectral and spatial techniques; ( ii ) identi® cation of all potentially signi® cant spectral clusters in the scene to be classi® ed; ( iii ) minimum distance classi® cation; and (iv) the use of spectral, spatial and large-scale pattern information in the progressive merging of the increasingly dissimilar clusters. The procedure was tested with high(Landsat Thematic Mapper (TM)) and medium(Advanced Very High Resolution Radiometer (AVHRR) 1 km composites) resolution data. It was found that the CPG yields classi® cation accuracies comparable to, or better than, current unsupervised classi® cation methods, is less sensitive to control parameters than a commonly used unsupervised classi® er, and works well with both TM and AVHRR data. The CPG requires only three parameters to be speci® ed at the outset, all specifying sizes of clusters that can be neglected at certain stages in the process. Although the procedure can be run automatically until the desired number of classes is reached, it has been designed to provide information to the analyst at the last stage so that ® nal cluster merging decisions can be made with the analyst’s input. It is concluded that the strategy on which the CPG is based provides an e€ ective approach to the classi® cation of remote sensing data. The CPG also appears to have a considerable capacity for data compression.