The middle part of the Lantau Island is selected as the experimental area in the study. The potential application of remote sensing and GIS are discussed and the relationship between landslide and environmental factors is analyzed. In the study, geological, geomorphological and erosion maps and DEM etc. are input to images, in which the information of surface objects are strengthened. Multi - original data Fusion images are created. The BP neural network classification is applied in the study area to recognizing regional natural landslides. The results show that the recognizing ability has been markedly improved. The method will provide an important scientific basis for the investigation and monitor of regional landslides.