\　A　compressive　light　field　camera　architecture　has　been　proposed　to　recover　light　fields　from　a　single　image.　This　technique　has　recently　gained　increasing　interest.　The　reconstruction　quality　of　light　field　depends　on　the　incoherence　of　a　projection　matrix　and　dictionary.　Compressive　light　field　acquisition　is　different　from　acquiring　traditional　signal　while　the　projection　matrix　has　specific　structural　features.　In　this　paper,　we　propose　a　new　design　for　the　mask　to　compute　the　optimized　projection　and　a　new　method　to　optimize　the　dictionary　for　compressive　light　field.　The　contribution　of　this　paper　comprises　of　twofold:　designing　the　mask　to　compute　the　optimized　projection　matrix　from　a　known　dictionary　of　compressive　light　field,　and　training　the　optimized　dictionary　when　the　projection　is　fixed.　Experimental　results　show　that　our　method　can　improve　the　reconstruction　quality　of　light　field　views　comparing　with　the　random　projection　mask.