Taxi　is　an　indispensable　part　of　the　urban　public　transportation　industry.　However,　the　existence　of　illegal　operating　behaviors　brings　security　risks　to　passengers　and　affects　the　city＇s　civilization.　The　paper　studies　the　detection　methods　of　taxi　drivers＇　smoking　behavior　during　operation　in　the　surveillance　video　environment.　Firstly,　the　Vibe　algorithm　is　used　to　detect　the　motion　foreground　of　the　video.　Then　the　Haar-Adaboost　algorithm　is　used　to　identify　the　taxi　and　extract　the　target　detection　area.　The　HSV　color　model　is　used　to　extract　the　color　characteristics　of　the　smoking　smoke,　and　separate　the　target　area　from　the　moving　objects　which　is　similar　to　smoking　smoke.　Finally,　combined　changes　in　shape　and　movement　characteristics　of　smoking　smoke　to　further　eliminate　the　interference　objects,　and　the　final　detection　results　are　obtained.　The　experimental　results　show　that　this　algorithm　has　high　real-time　and　accuracy,　which　is　beneficial　to　off-site　law　enforcement　in　intelligent　traffic　management.