Building 3D Change Detection Analysis Based on Spatial Voxel Expression and Image Target Recognition
Aiming at the problems of low detection efficiency,false alarm and missing detection of illegal buildings in complex urban and rural areas,in this paper,a 3D change detection method for buildings suitable for dense matching point clouds of UAV images is proposed by combining 3D point cloud spacial voxel expression and 2D image target recognition.First,the two-phase UAV dense matching point clouds are pre-processed,including point cloud registration and ground point filtering;then,an octree is constructed for the point cloud,and voxels are compared layer by layer to obtain the changed point cloud.Finally,the changed point cloud is clustered and projected one by one according to the orthographic perspective to generate 2D images.The ConvNext network is used to identify buildings to obtain building change detection results.The experiments evaluate the inspection effect at the object level,achie-ving 100%detection completeness with an accuracy rate of 87.18%.The results show that the proposed method can effectively im-prove the efficiency of illegal building detection and meet actual production needs.
densely matched point cloudsillegal buildings3D change detectionspatial voxeltarget recognition
万方数据
维普
