A　new　method　combining　CANDIDE　model　with　FAPs　space　parameters　is　proposed　to　track　face　features　and　extract　face　animation　parameters.　This　paper　trains　the　relationship　of　face　animation　parameters　and　different　face　images,　and　directly　doing　face　animation　parameters　extraction　and　tracking.　Experimental　results　show　that　the　algorithm　is　robust　and　accurate.