Implementation of Background Subtraction for Counting Vehicle Using Mixture of Gaussians with ROI Optimization

There is an imbalance between the ratio of the number of vehicles of 11% and the addition of new roads or road extensions of 0,01%, especially in Jakarta, Indonesia, which is often an issue that causes traffic problems, one of them is traffic jam. This paper discusses an implementation of a video surveillance system-based method to monitor traffic conditions such as detection, tracking and counting of vehicles in the form of information technology in the form of system simulation using a computer.The objective of this research is the implementation of a video surveillance based system that can detect, track and count the number of vehicles using an image processing method approach. The approach used in this research is Mixture of Gaussians (MOG2) for background subtraction with optimization of Region of Interests (ROI). There are four stages in this method, namely pre-processing, vehicle tracking, vehicle counting, and ROI optimization. The results were obtained in the form of accuracy which is divided into two conditions, namely in the morning and in the daytime. For accuracy, this system has a capability of 86% in the morning and 94,1% in the daytime with each video duration of 30 seconds. This system simulation can be used as a reference for traffic-related bureaus to help manipulate traffic.