Satellite Imagery and Exogenous Data Integration by Neural Network in AutomationLand-Cover Classification

摘要

Certainly data integration for land-cover classification requires a non-linearsystem to associate satellite imagery with exogenous imagery. In this study we present some results of a Neural Network based methodology to provide land-cover classifications. Two approaches are investigated: (1) The Monolithic integration: all required registered images are the inputs of only one Back-Error Propagation (BEP) network. The network is trained on purpose to get the final classification. (2) The class-distributed integration: for each class a specific network learns from all satellite imageries its class characteristics. In both approaches, topographic mapping is taken into account as exogenous data.