This　article　presents　a　novel　vehicle　edge　detection　algorithm　based　on　Canny　operator　in　nighttime　environment.　Firstly,　spatial　homomorphism　filtering　is　used　to　eliminate　the　illuminating　influence;　then　an　improved　self-adaptive　fuzzy　enhancement　Canny　operator　is　used　to　enhance　the　vehicle　edge;　lastly　the　maximum　classification　square　error　method　is　used　to　calculate　the　dual　threshold,　performing　the　vehicle　edge　detection.　Experiments　indicate　that　this　method　eliminates　the　interference　from　the　auto　lamps,　well　applicable　for　the　vehicle　edge　detection　in　the　night-time　environment.　©2010　IEEE.