In　order　to　extract　all　the　billiard　objects　from　Snooker　videos,　an　effective　algorithm　of　the　billiard　segmentation　is　proposed　in　this　paper.　This　method　is　divided　into　the　follow　two　steps:　first,　we　use　an　automatic　segmentation　method　of　local　peak　edges　to　extract　the　Snooker　table,　this　step　is　mainly　served　for　the　image　preprocessing;　and　then　is　the　detection　of　video　objects.　Through　image　preprocessing,　morphological　processing　and　clustering,　we　can　achieve　the　candidate　areas　of　billiard　balls.　With　the　color　space　conversion　between　HSV　and　RGB　and　histogram　equalization　in　digital　image　processing,　the　billiard　areas　of　accurate　detection　are　derived　from　the　characteristics　of　the　candidate　areas.　Experimental　results　are　represented　in　three　dimensional　reconstruction　scenes.　©2010　IEEE.