With the continuous advancement of 3D real scene China construction,city level 3D real scene modeling serves as crucial data infrastructure,where 3D building models constitute essential components.Current methods relying on single data sources for urban building reconstruction often suffer from defects such as voids,deformation,and texture stretching,leading to incomplete modeling.To address these issues,an integrated 3D modeling method combining UAV oblique photogrammetry images with airborne LiDAR point cloud data through registration and fusion is proposed in this paper.Initially,both LiDAR point clouds and UAV imagery are acquired through aerial surveys,with UAV point clouds generated via aerial triangulation.Subsequently,a two stage point cloud registration approach is implemented:coarse alignment followed by refined registration using a KD tree optimized Iterative Closest Point (ICP) algorithm,achieving effective point cloud fusion for complete 3D reconstruction.Experimental validation through the 3D real scene construction project in Mengzi City,Honghe Prefecture,Yunnan Province demonstrates that the KD tree enhanced ICP algorithm significantly improves registration efficiency and accuracy.The fused 3D models exhibit enhanced completeness and refinement in architectural details,effectively resolving issues of model voids,deformation,and texture artifacts.