3D　laser　scanning　can　obtain　a　lot　of　point　cloud　data,　the　display　speed　of　which　is　directly　affected　by　their　structure.　For　the　low　efficiency　in　managing　and　visualizing　the　point　cloud　data,　this　paper　proposed　a　spatial　index　structure　mixing　the　improved　octree　and　the　leaf　node　of　k-d　tree,　as　well　as　a　level　of　detail　(LOD)　model.　Globally,　quick　indexing　and　management　can　be　realized　by　the　octree　model;　locally,　efficient　query　and　display　can　be　realized　by　the　k-d　tree　constructed　in　memory.　This　mixed　data　model　was　adopted　to　structure,　establish　spatial　index,　and　construct　point　cloud　data　by　LOD,　thereby　realizing　the　index　and　visualization　of　point　cloud　data.