Improved Accuracy of Vehicle Counter for Real-Time Traffic Monitoring System

• Main Authors: Setiadi De Rosal Ignatius Moses, Fratama Rizki Ramadhan, Partiningsih Nurul Diyah Ayu
• Format: Article
• Language: English
• Published: Sciendo 2020-04-01
• Series: Transport and Telecommunication
• Subjects: traffic monitoring; background subtraction; truncate threshold; real-time tracking; vehicle detection
• Online Access: https://doi.org/10.2478/ttj-2020-0010

This research proposes a background subtraction method with the truncate threshold to improve the accuracy of vehicle detection and tracking in real-time video streams. In previous research, vehicle detection accuracy still needs to be optimized, so it needed to be improved. In the vehicle detection method, there are several parts that greatly affect, one of which is the thresholding technique. Different thresholding methods can affect the results of the background and foreground separation. Based on the results of testing the proposed method can improve accuracy by more than 20% compared to the previous method. The thresholding method has a considerable influence on the final result of vehicle object detection. The results of the average accuracy of the three types of time, i.e. morning, daytime, and afternoon reached 96.01%. These results indicate that the vehicle counting accuracy is very satisfying, moreover, the method has also been implemented in a real way and can run smoothly.