Three-dimensional　laser　scanning　can　be　used　to　collect　the　3D　coordinate　information　on　the　object　surface　to　create　3D　point　clouds,　which　can　reflect　the　characteristics　of　the　target　surface,　facilitating　the　establishment　of　terrain　model.　Since　point　clouds　lacunarity　is　usually　caused　by　occlusion　of　ground　features　during　the　vehicle-borne　laser　scanning,　the　existing　modeling　algorithms　are　not　fully　suitable　for　vehicle-borne　laser　point　clouds.　In　this　paper,　a　modeling　algorithm　was　designed　to　be　different　from　the　previous　ones,　in　which　the　model　was　built　by　constructing　Delaunay　triangulation.　Instead,　a　bounding　box　algorithm　was　adopted　for　the　modeling　of　vehicle-borne　laser　point　clouds,　which　were　structured　in　an　octree.　This　could　not　only　fill　the　point　cloud　holes　found　in　modeling,　but　also　improve　the　storage　efficiency　of　the　point　clouds.