Abstract: An automatic landuse change detection approach combining high spatial resolution remotely sensed image with GIS data is proposed. Mahalanobis distance, Support Vector Machine (SVM) and Neural Network (NNT) are used respectively for fuzzy classification. To improve the overall accuracy of classification,  a fuzzy decision fusion classification algorithm is designed to combine each fuzzy classification result from the above methods.  Based on these accurate classification results and the GIS data in historical period, each classification in the corresponding polygon is calculated. The changed area is automatically detected by comparing the calculated land status ratio with its historical status. QuickBird images and GIS data are taken for an experiment. Polygon which exhibits big change of land status ratio can be automatically detected and judged accurately; while polygon which has little change may be subject to false judgment. Experimental results proved that fusion classification results are more accurate than the individual classification result.