The　proliferation　of　video　content　makes　video　similarity　detection　an　indispensable　tool　in　video　management,　searching,　and　navigation.　In　this　paper,　a　novel　video　retrieval　algorithm　targeted　on　detecting　given　video　clips　from　TV　programs　is　proposed.　The　algorithm　has　the　following　two　advantages.　Firstly,　the　video　spatiotemporal　feature　used　in　this　algorithm　has　many　favorable　features,　such　as　easily　and　uniquely　extraction,　and　drastically　video　information　reduction.　Secondly,　the　algorithm　is　robust　against　the　whole　offset　of　color　or　brightness　and　abrupt　intense　disturbance,　which　are　frequently　seen　on　TV　programs.　Grads　comparison　method　and　Exception　Factor　were　introduced　to　enhance　algorithm　robustness.　Experimental　results　demonstrated　the　effectiveness　of　the　algorithm.