A　recognition　algorithm　for　hand　vein　based　on　speeded-up　robust　features(SURF)　was　proposed.　Firstly,　the　region　of　interest　(ROI)　of　the　hand　vein　image　is　obtained　through　image　preprocessing.　The　local　SURF　features　of　test　sample　and　register　samples　are　extracted　and　matched　based　on　Euclid　distance,　and　then　the　mismatching　pairs　are　rejected.　Finally,　the　matching　rate　is　calculated　as　the　similarity　between　registered　sample　and　test　sample　to　realize　personal　identity　recognition.　The　recognition　performance　was　evaluated　in　verification　mode　using　TJU　hand　vein　image　database,　the　equal　error　rate　(EER)　is　0.07%　and　average　recognition　time　is　0.153　s.　Experimental　results　show　that　the　proposed　algorithm　is　able　to　realize　hand　vein　recognition　reliably　and　quickly.