Facial　expression　recognition　is　widely　used　in　linguistics,　medical　care,　service　and　so　on..　But　still　there　exists　the　problem　of　accuracy..　To　solve　this　problem,,　this　paper　presents　an　extraction　method　based　on　the　flow　characteristics　of　facial　motion　unit　for　the　facial　expression　recognition.　The　facial　action　coding　system　(FACS)　is　first　applied　to　extract　facial　motion　unit,　and　then　optional　flow　method　for　these　areas　is　used　to　extract　eigenvalue　to　constitute　the　signature　sequence,　finally　categorization　based　on　random　forests　is　applied　to　classify　and　recognize　the　smiley　facial　expression.　Experimental　results　validate　the　effectiveness　of　the　proposed　scheme.