In　order　to　improve　driving　safety,　this　paper　proposes　a　Vehicle　Identification　approach　by　two-dimensional　laser　radar.　The　approach　can　detect　obstacles　in　front　of　vehicle　and　determine　whether　the　obstacles　are　vehicle.　In　order　to　eliminate　the　effect　of　system　errors　generated　in　the　process　of　Laser　radar　data　collection,　Median　filtering　method　is　used　in　radar　data　processing.　In　order　to　determine　the　number　and　position　of　obstacles　in　front　of　vehicle,　CURE　clustering　algorithm　is　proposed　to　clustering　radar　data.　The　clustering　algorithm　can　solve　the　problem　that　centroid-based　clustering　approach　is　not　suitable　for　non-spherical　or　arbitrary　shaped　clusters.　Improved　the　CURE　algorithm,　the　improved　algorithm　is　more　suitable　for　clustering　the　data　of　two-dimensional　laser　radar.　The　target　of　vehicle　identification　is　achieved　by　calculating　the　width　and　length　of　each　cluster.　The　conclusion　verified　by　experiments　is　this　approach　can　calculate　the　width　and　length　of　obstacle　more　precisely　and　recognize　vehicle　accurately.　©　2011　IEEE.