In　this　paper　we　address　the　problem　of　how　to　identify　real　4K　video　contents　from　pseudo　ones　which　were　generated　using　complicated　image　post-processing　technologies　such　as　upsampling　or　super-resolution　(SR).　Two　individual　sets　of　22　4K　video　sequences　and　6　popular　SR　algorithms　were　collected　for　conducting　this　research.　Despite　the　superior　performance　of　current　SR　methods,　they　still　not　reach　the　perfect　level　since　SR　image　reconstruction　is　a　severely　ill-posed　problem　and　details　are　hard　to　be　entirely　recovered.　This　implies　the　reconstructed　images　based　on　SR　methods　do　not　conform　to　the　natural　scene　statistics　(NSS)　model.　According　to　this,　we　explore　the　difference　of　the　locally　normalized　luminances’　distributions　between　real　4K　video　sequences　and　SR-generated　pseudo　4K　video　sequences.　It　was　shown　by　extensive　tests　that　simple　statistical　features　can　be　fused　to　form　a　good　classifier　for　separating　real　4K　video　contents　from　pseudo　versions.　Experimental　results　demonstrate　that　the　proposed　classier　consumes　less　than　a　quarter　of　a　second　for　a　4K　video　frame　and　the　accuracy　of　classification　is　beyond　90%.　abstract　environment.　©　2018,　Springer　Nature　Singapore　Pte　Ltd.