The　3D　face　recognition　attracts　more　and　more　attention　recently,　because　of　its　insensitivity　to　the　variance　of　illumination　and　pose.　In　this　paper,　a　novel　self-adaptive　3D　face　recognition　algorithm　based　on　face　feature　division　is　proposed.　This　algorithm　first　divides　the　3D　face　into　some　block　regions　based　on　face　features,　and　then,　four　face　surface　characters　are　extracted　for　all　the　block　regions　to　represent　the　3D　face,　and　novel　self-adaptive　3D　face　recognition　algorithm　is　proposed　to　balance　the　contributions　of　all　the　block　regions.　After　that,　a　linear　weighted　strategy　is　adopted　to　hybrid　the　four　characters　and　boost　the　recognition　rate　further.　Finally,　we　test　our　algorithm　on　BJUT-3D　face　database　and　concluded　that　the　performance　of　our　algorithm　and　fusion　strategy　is　satisfying.