In　this　paper　we　describe　a　video　surveillance　system　that　is　not　only　able　to　detect　human　face　but　also　to　track　it.　The　given　algorithm　is　composed　of　three　steps.　Firstly,　an　improved　skin　detection　method　is　used　to　acquire　promising　face　regions　of　the　image.　Then　it　uses　the　continuous　Adaboost　algorithm　to　locate　faces.　The　face　tracking　algorithm　is　the　mean　shift　tacking　method　which　is　based　on　the　histograms　of　moving　objects　in　consecutive　image　frames.　The　experiment　results　show　that　the　method　is　translation,　rotation,　scale　invariant　and　computationally　efficient.　The　results　obtained　for　test　video　sequences　are　very　encouraging.　It　provides　the　ability　to　detect　and　track　the　face　in　video　surveillance　with　complex　scenes.　1548-7741/Copyright　©　2008　Binary　Information　Press　October　2008.