This　paper　presents　a　lane　detection　method　to　screen　data　using　the　vanishing　point　according　to　the　perspective　feature　of　the　camera.　The　line　data　is　obtained　after　Canny　and　Hough　transform　of　the　raw　image.　The　filter　conditions　are　created　according　to　the　vanishing　point　and　other　location　features.　The　algorithm　saves　the　detected　lane　and　vanishing　points　in　near　history.　The　algorithm　clusters　and　integrates　to　determine　the　detection　output　according　to　the　historical　data.　Finally,　according　to　the　output,　a　new　vanishing　point　is　fitted　for　the　next　circuit.　The　experiments　indicate　that　the　algorithm　detects　the　lane　accurately　and　is　of　parameters　self-adaptability　and　robustness,　while　improving　the　utilization　of　the　internal　data.