Identifying　vital　nodes　is　crucial　in　researching　the　structures　and　evolution　of　complex　networks.　Most　existing　link　prediction　methods　utilize　node　degree　as　the　measure　of　node　importance.　But　degree　is　less　accurate　in　evaluating　the　importance　of　nodes　since　it　exploit　very　limited　information.　Therefore,　we　introduce　node　centrality　to　identify　vital　nodes.　This　paper　proposes　a　link　prediction　method　based　on　node　centrality　to　improve　accuracy,　which　can　distinguish　the　endpoint　influence　and　path　connectivity.　We　reveal　that　closeness　centrality　describe　the　endpoint　influence　better　than　degree　and　betweenness　centrality.　and　betweenness　centrality　quantifies　the　path　connectivity　best.