Point Cloud Extraction of Building Roof and Facade Based on CSF and CANUPO
Oblique aerial images have the characteristics of high overlap rate,visible building facades and less occlusion,which can effec-tively extract building information.In this study,tilting aerial images will be processed to generate dense point clouds,and ground points and non-ground points will be filtered out by cloth simulation algorithm(CSF),and non-ground points will be de-noised.The CANUPO algo-rithm is used to classify non-ground points into buildings,vegetation and others.Then the normal calculation of the building points,according to the normal results to extract the building roof points and facade points,the extraction accuracy can reach about 80％of manual extraction,this method can simply and effectively provide users with the initial information of the building roof and facade in a given area.
oblique aerial imagedense point cloudcloth simulation(CSF)CANUPOnormal estimation
万方数据
维普
