In　art,　dynamic　sculptures　have　long　faced　the　dilemma　of　poor　interaction　with　the　viewer.　Faced　with　this　situation,　this　research　has　developed　an　intelligent　interaction　system　for　dynamic　sculptures　that　uses　advanced　computer　vision　technology　to　accurately　detect　every　movement　of　the　viewer　and　collect　information　on　their　location.　The　system　analyzes　the　collected　data　quickly　in　real-Time　and　skillfully　translates　them　into　commands　for　manipulating　the　dynamic　sculpture,　thus　displaying　the　responses　smoothly.　Experimental　data　show　that　the　success　rate　of　the　system＇s　interactive　mode　reaches　98.90%.　The　success　rate　of　the　system＇s　interactive　mode　can　reach　98.90%,　greatly　enhancing　the　audience＇s　sense　of　participation　and　strongly　confirming　the　effectiveness　and　unmatched　superiority　of　computer　vision　technology　applied　to　the　interactive　dynamic　sculpture　system.　©　2015　SPIE.