A　novel　skin　color　detection　method　in　JPEG　compressed　domain　has　been　proposed.　Color　and　texture　features　of　the　image　blocks　are　extracted　from　the　entropy　decoded　DCT　coefficients　firstly.　Then,　data　mining　method　is　applied　to　set　up　the　skin　color　model　to　describe　the　relationship　between　the　image　block　features　and　the　skin　detection　results,　and　the　initial　skin　image　blocks　are　detected　based　on　the　model.　The　skin　color　regions　are　finally　segmented　using　region　growing　method.　Experimental　results　show　that,　compared　with　the　SPM　(Skin　Probability　Map)　skin　color　detection　algorithm　in　pixel　domain,　the　proposed　method　can　achieve　higher　detection　accuracy　and　faster　detection　speed.