In　this　paper　we　present　a　detection　algorithm　of　road　cracks　in　several　directions　and　each　layer　for　pavement　images　in　contourlet　domain.　Firstly,　a　pavement　image　is　converted　to　a　grey-scale　image　which　is　decomposed　by　using　contourlet　transform.　Then　directionality　and　anisotropy　are　used　to　enhance　the　singular　characteristic　of　the　image　by　expanding　all　scale　detail　coefficients　to　the　same　size　and　combining　the　coefficients.　The　new　approach　avoids　unreasonable　evaluation　result　of　classical　unified　crack　index　pavement　distress　evaluation　method　based　on　image　tile.　Experimental　results　show　that　the　algorithm　proposed　in　this　paper　can　detect　pavement　cracks　effectively　which　is　not　affected　by　noise.