Most　recognition　algorithms　are　for　2D　data　with　strict　restrictions,　these　algorithms　are　easily　affected　by　pose,　illumination　and　other　factors.　The　reason　is　due　to　the　shortage　of　information.　To　resolve　this　problem,　we　present　a　face　recognition　algorithm　based　on　fusing　2D　and　3D　information.　Being　different　from　other　algorithms,　the　input　of　our　algorithm　is　one　single　2D　gray-scale　image　and　3D　information　is　provided　by　reconstructed　the　3D　model.　For　the　2D　image,　we　choose　LBP　feature　as　face　representation　feature.　For　the　3D　model,　we　define　54　feature　points,　calculate　geodesic　distances　between　the　nose　tip　and　other　feature　points　as　3D　feature.　A　weighted　sum　of　the　2D　and　3D　scores　is　used　to　deliver　the　fusion　process　and　the　weights　are　determined　based　on　Fisher　Linear　Discriminant　Analysis.　Finally,　the　presented　algorithm　is　tested　on　CAS-PEAL-R1　face　database　with　illumination.