In　order　to　increase　the　performance　of　human　face　orientation　nested　in　individual　AmI　(Ambient　intelligence)　system　and　to　reflect　the　concept　of　AmI　serving　anywhere　and　hidden　computing,　an　algorithm　of　face　orientation　based　on　wavelet　is　proposed　here.　First,　the　position　information　of　face　was　obtained　from　a　lot　of　real-time　acquired　images　that　were　preprocessed　by　skin-model.　Second,　using　wavelet　analysis　located　the　geometrical　shape　of　hair　area,　and　then　combining　prior　knowledge　gained　the　hair　geometric　probability　model.　Finally,　this　model　was　applied　to　the　AmI　system　of　real-time　position　and　tracing　so　as　to　implement　the　individual　video　service　for　the　AmI　users.　The　results　show　that　the　proposed　algorithm　is　capable　of　locating　the　human　face　orientation　in　the　AMI　system　with　good　performance　of　real-time　and　high　accuracy　of　positioning.