This　paper　presents　an　effective　scene-detection　approach　for　unnamed　news　video.　Firstly,　we　use　shot　detection　based　on　the　hand-craft　feature　to　divide　the　unnamed　news　video　into　shots.　Secondly,　convolutional　neural　network　and　characteristic　of　host　scene　were　combined　to　identify　the　news　channel.　Finally,　we　take　advantage　of　the　continuity　of　news　video　and　propose　a　new　scene　detection　method.　Our　team　has　tested　the　proposed　method　on　a　variety　of　videos　including　videos　from　CCTV4,　CCTV13,　Beijing　News,　Anhui　News　and　Shanghai　News.　Comparative　analysis　shows　that　our　method　outperforms　the　competitor　approach　using　similar　experimental　settings,　which　proves　the　effectiveness　of　our　method　on　news　video　scene　detection　©　2017　IEEE.