In　order　to　reduce　the　computational　complexity　of　H.264　frame-skipping　transcoding,　an　support　vector　machine　(SVM)　technology　is　employed　in　inter　mode　decision　process,　and　a　SKIP　mode　decision　algorithm　is　proposed.　Meanwhile,　an　early-termination　strategy　is　used　based　on　the　temporal　correlation　between　macro-blocks,　and　improbable　candidate　modes　can　be　eliminated.　In　addition,　an　adaptive　motion　vector　reconstruction　and　motion　compensation　algorithm　is　applied　for　sequences　with　different　motions　by　measuring　the　motion　activity　of　macroblocks,　and　the　computational　redundancy　can　be　removed　in　case　that　the　best　matching　motion　vector　can　be　found　before　motion　search.　Experimental　results　show　that　the　coding　time　is　reduced　by　an　average　of　84%　dramatically　while　maintains　the　video　image　of　high　quality.　Compared　with　the　existing　approaches,　this　algorithm　is　more　robust　and　applicable.