This study presents an implementation of an intelligent traffic light control system for a crossroads based on combination of image processing and fuzzy controller. After certain seconds of red light, traffic densities are estimated from captured images. Based on this estimated result, a fuzzy controller determines the length of the next green light period in order to generate longer period for higher density road, vice versa. | T¤p ch½ Tin håc v i·u khiºn håc, , (2012), 373 383 PH T TRI N GI I THU T I U KHI N N GIAO THÆNG THÆNG MINH NGÆ L NGÅC B CH, NGUY N CH NGÆN Tr÷íng ¤i håc C¦n Thì Tóm t t. B i b¡o · xu§t mæ h¼nh c i °t h» thèng ±n giao thæng ¡p döng cho mët ng¢ t÷ düa tr¶n sü k¸t hñp cõa kÿ thuªt xû lþ £nh v i·u khiºn mí. Sau khi ±n ä ÷ñc bªt v i gi¥y, h¼nh £nh s³ ÷ñc chöp v ÷îc l÷ñng mªt ë xe l÷u thæng. Düa tr¶n k¸t qu£ ÷îc l÷ñng n y, bë i·u khiºn mí s³ quy¸t ành thíi gian ±n xanh cho chu ký ti¸p theo, vîi möc ti¶u l m cho tuy¸n ÷íng câ mªt ë xe l÷u thæng cao hìn th¼ thíi gian cõa chu ký ±n xanh s³ d i hìn v ng÷ñc l¤i. Ch÷ìng tr¼nh mæ phäng ÷ñc x¥y düng º ¡nh gi¡ mæ h¼nh · xu§t công nh÷ so s¡nh vîi h» thèng ho¤t ëng theo cì ch¸ ành thíi phê bi¸n hi»n nay. K¸t qu£ mæ phäng ¢ chùng minh ÷ñc thíi gian ±n xanh phö thuëc v o mªt ë xe tr¶n ÷íng, xe c ng nhi·u, thíi gian c ng d i v ng÷ñc l¤i. Khi mªt ë xe giúa 2 tuy¸n ÷íng ch¶ch l»ch nhau l 5 l¦n th¼ têng thíi gian chí cõa mæ h¼nh ho¤t ëng theo cì ch¸ thæng minh gi£m tø 15 ¸n 20 l¦n so vîi cì ch¸ ành thíi. Abstract. This study presents an implementation of an intelligent traffic light control system for a crossroads based on combination of image processing and fuzzy controller. After certain seconds of red light, traffic densities are estimated from captured images. Based on this estimated result, a fuzzy controller determines the length of the next green light period in order to generate longer period for higher density road, vice versa. A software was developed to evaluate the suggestion algorithm, and to compare it with traditional timer-based traffic light system. Simulation results indicate that the road with higher traffic flow has a longer green light period, and vice versa. When the traffics densities of 2 roads are 5 times different, the vehicle red-light waiting times of the suggested system are reduced from 15 to 20 times better than the traditional .
