In　machine　vision　navigation,　quickly　and　accurately　recognizing　the　direction　of　curve　lanes　is　an　important　step.　However,　many　existing　methods　to　determine　directions　of　curve　lanes　have　some　problems　with　accuracy　and　real-time　performance.　Aiming　to　solve　these　two　problems,　this　paper　proposed　a　method　based　on　Hough　Transform　to　determine　the　curved　lanes　direction.　Our　method　first　recognizes　two　straight　lines　respectively　from　left　side　and　right　side　of　curve　lane　and　then　combine　dynamic　sensitive　curve　areas(DSCA)　with　vanish　point　of　two　detected　straight　lines　to　obtain　characteristic　value　of　the　curve　lane　direction.　Finally　the　direction　of　curve　lane　can　be　determined　by　the　rule　of　comparing　with　two　thresholds　of　direction.　The　method　was　verified　by　experiment,　and　it　can　under　the　premise　in　high　accuracy,　and　improve　the　real-time　performance.