Driver　fatigue　is　an　important　cause　of　combination-unit　traffic　crashes　and　eyes　are　an　important　factor　to　embody　driver＇s　fatigue　degree.　At　present,　almost　all　of　driver＇s　eyes　detecting　and　tracking　systems　are　based　on　one　camera.　Such　system　of　detecting　and　tracking　eyes＇　features　will　give　very　lower　accuracy.　So　a　method　of　driver＇s　eyes　detection　and　tracking　system　using　two　cameras　is　proposed　in　this　paper.　It　is　based　on　a　controllable　camera　and　a　fixed　camera　and　the　two　cameras　can　communicate　through　serial　port.　The　main　principle　is　to　collect　the　monitoring　video　of　the　cabin　through　camera　A　(shoot　driver＇s　upper　body)　at　first,　then　locate　the　driver＇s　face　using　rapid　image　processing　algorithm　in　camera　A.　After　that　we　control　camera　B　(just　shoot　driver＇s　face)　through　serial　port　according　to　face　position　in　camera　A　and　make　it　shoot　driver＇s　face　all　the　time.　At　last,　a　new　method　which　is　based　on　combined　Adaboost　algorithm,　particle　filter　and　prior　knowledge　was　applied　to　track　driver＇s　eyes　in　camera　B.　©　2010　ASCE.