In　recent　years,　finger　vein　recognition　has　been　favored　by　more　and　more　researchers　because　of　its　high　recognition　accuracy,　security　and　convenience　of　collection.　The　rotation　of　the　finger　will　reduce　the　recognition　performance.　This　paper　first　correction　the　collected　images　through　the　smallest　circumscribed　rectangle,　then　extracts　the　region　of　interest　according　to　the　location　of　finger　joints,　and　extracts　vein　features　based　on　Niblack　algorithm.　Finally,　the　intersection　points　and　endpoints　of　the　veins　are　identified,　and　an　modified　Hausdorff　distance　algorithm　(MHD)　is　used　to　identify.　The　experiment　shows　that　the　rotation　average　time　and　the　extraction　time　of　the　venous　feature　of　each　picture　are　8ms　and　146ms,　respectively.　The　accuracy　of　the　non　image　rotation　correction　is　94.12%,　and　the　accuracy　of　the　image　rotation　correction　is　97.21%,　and　the　algorithm　is　robust　to　the　rotation　angle.　It　can　be　concluded　that　the　algorithm　has　a　high　advantage　in　running　speed　and　matching　precision.　©　2018　IEEE.