With　an　increasing　number　of　video　acquisition　devices,　the　amount　of　video　data　dramatically　grows.　Key　frame　extraction　can　immensely　decrease　the　amount　of　data　required　in　the　field　of　video　analysis.　In　this　paper,　we　present　a　multi-view　(descriptors)　key　frame　extraction　method　based　on　curvature　point　detection　for　videos　of　various　genre.　This　method　automatically　estimates　the　weights　that　indicate　the　importance　of　each　descriptor　for　shots.　The　weight　of　each　descriptor　is　used　to　form　a　composite　frame　difference　measure　as　the　weighted　sum　of　all　the　frame　difference　measures　between　the　consecutive　frames　corresponding　to　the　individual　descriptor.　The　final　frame　differences　construct　a　curve　of　cumulative　frame　differences　within　which　the　high　curvature　points　are　detected.　The　frames　corresponding　to　the　midpoints　between　each　pair　of　consecutive　high　curvature　points　are　selected　as　the　key　frames.　The　proposed　method　is　performed　on　publicly　available　video　datasets　and　experiment　results　demonstrate　that　compared　with　existing　related　works　our　method　yields　better　results　for　key　frames　extraction.　©　2018　Indian　Pulp　and　Paper　Technical　Association.　All　rights　reserved.