應用彩色影像辨識技術於交通車流參數之蒐集與分析

Abstract

本研究以一般個人電腦(PC)及攝影機(Hi-8)等硬體設備，透過彩色影像辨 識技巧，以Visual Basic 4.0軟體設計一套電腦程式來蒐集並分析瞬時性 交通車流參數。而電腦系統程式共包含了影像數位化、虛擬偵測器、背景 比對與交通參數析出等四個程式模組。利用本偵 測系統可求得之交通 車流參數計有不同車種(大型車、小型車與機車)之車輛數及總車輛數、車 頭距(Time Headway)、車間距(Time Gap)、平均車頭距、平均車間距、佔 有率(Occupancy)、速率(Speed)、流量(Flow)、平均速率(Space Mean Speed)、密度(Density)等十三種。經由道路實測結果，在高速公路(非干 擾性車流狀況下)之平均偵測準確率日間為97.6%、夜間為94.5%；市區(汽 、機車干擾性車流狀況下)之平均偵測準確率日間為88.7%，夜間則未觀測 。 This thesis attempts to develop a traffic flow image detection system (TFIDS) to analyze instantaneous traffic flow characteristic data using color image processing techniques . The system is composed of four modular computer programs including image digital module ,virtual detector module , image processing module ,and traffic flow characteristic data calculating module . The TFIDS can collect and analyze such data as time headway , average headway , time gap , average gap , occupancy , speed , flow , space mean speed , density ,and the numbersof trucks , autos ,and motorcycles . To demonstrate the applicability of thissystem , both uninterrupted and interrupted traffic flow environments are tested .The testing results show that accuracy can reach 97.6% in the daytime and 94.5%in the nighttime on the freeway mainline section and 88.7% in the daytime on the urban intersection approach .