TRAFFIC CONGESTION ESTIMATION USING VIDEO WITHOUT VEHICLE TRACKING

• Main Author: Anggriawan Siswoyo, Antony
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/45141
• Institution: Institut Teknologi Bandung

The congestion level of traffic is one of many factors that usually used as consideration material of the community to travel such as trip to the offuce, school, vacation and many more. These informations are used by the community to determine the path they will take to get to their destination. To obtain and disseminate these informations, the government uses a CCTV (Closed-Circuit Television) system to conduct surveilance on the traffic, but the monitoring system used by the government cannot be applied to urban traffic because of the habits of drivers are irregular. For this very reason, another system is needed to monitor traffic condition in the city. To build the system, we must first examine how to process the image that come from the camera to become information of traffic level congestion. Based on existing studies, one of the algorithm commonly uses in this kind of system is the Expectation Maximization algorithm. This algorithm serves to classify processed data obtained from image or video that come from observed traffic. In this system, the use of Expectation Maximization (EM) algorithm is collaborated with the Image Thresholding method, where this method procceses image or data video so it can produce the value needed by the EM algorithm. In this study, the traffic density monitoring system , using the EM algorithm, will be compared of result with the traffic video data that has varying density level. Also, an experiment will be conducted on the Image Thresholding method that will be used for accuracy of the value that will be processed by the EM algorithm. Based on the research, the EM successfully detected the density of the traffic where in this study the density of the traffic was separated into 2 levels, “crowded” and “smooth”. The use of local Image Thresholding method is the most accurate method to process video data in urban traffic.