In　modern　society,　with　the　economic　conditions　getting　better,　more　and　more　vehicles　are　produced.　So　many　vehicles　bring　about　the　rapid　development　of　traffic　transportation　and　the　traffic　accidents　happen　frequently.　With　the　fast　development　of　computer　technology,　more　and　more　interests　are　focused　on　vision　navigation　technology　in　Intelligent　Vehicle　System.　At　the　same　time,　lane　detection　is　an　important　component　of　intelligent　vehicle　vision　navigation　system.　In　order　to　improve　the　real-time　detection,　an　improved　lane　boundaries　detection　based　on　dynamic　regions　of　interest　is　presented.　The　approach　is　an　essential　tasking　both　autonomous　lane　vehicles　research　and　active　safety　system　development.　Lane　detection　is,　however,　still　a　challenging　issue　due　to　the　complexity　of　the　real　road　scenes.　Our　approach　takes　advantage　of　the　image　preprocessing　with　different　processing　on　the　images　with　different　illumination,　ROI　decision　algorithm　and　lane　detection　based　on　the　vanishing　point　and　Hough　transform.　The　achieved　results　reveal　that　the　Hough　transform　with　the　dynamic　ROI　algorithm　and　vanishing　point　method　is　more　effective.