An Approach based on Adaptive Decision Tree for Land Cover Change Prediction in Satellite Images
Résumé
Decision tree(DT)predictionalgorithmshavesignificantpotentialforremotesensingdataprediction.This
paper presentsanadvancedapproachforland-coverchangepredictioninremote-sensingimagery.Several
methods fordecisiontreechangepredictionhavebeenconsidered:probabilisticDT,beliefDT,fuzzyDT,and
possibilistic DT.TheaimofthisstudyistoprovideanapproachbasedonadaptiveDTtopredictlandcover
changes andtotakeintoaccountseveraltypesofimperfectionrelatedtosatelliteimagessuchas:uncertainty,
imprecision, vagueness,conflict,ambiguity,etc.Theproposedapproachappliesanartificialneuralnetwork
(ANN) modeltochoosetheappropriategainformulatobeappliedoneachDTnode.Theconsideredapproach
is validatedusingsatelliteimagesrepresentingtheSaint-Paulregion,communeofReunionIsland.Results
showgoodperformancesoftheproposedframeworkinpredictingchangefortheurbanzone.