Open AccessEstimation of Vehicle Speed at Night Based on Rear Lights
Author(s) -
Budi Setiyono,
D. Ratna,
Muhammad Fadli Ilmi,
I. G. N. R. Usadha
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1752/1/012087
Subject(s) - computer vision , computer science , frame (networking) , artificial intelligence , position (finance) , economics , telecommunications , finance
All Speed violations are common things. One way to prevent it is detecting the vehicle and estimating the speed of vehicle that is running. To detect the vehicle, we can use the vehicle lights. The rear lights of the vehicle could be detected easier than the front lights of the vehicle because the color of the rear lights was different than the common lights on the road at night. This paper focuses on estimating the speed of a vehicle that runs at night based on the rear lights of a vehicle. The first step is calibrating the camera used and extracting the video into frames. Then we tracked the vehicle position in each frame in the Region of Interest (ROI). The next step is bounding and labelling each detected vehicle. After that, the conversion from the camera angle and the camera height become the actual distance in each frame. The final step is to find the movement of the vehicle from one frame to another using Euclidean distance to determine the speed of the vehicle. This research results in high estimation accuracy when in the ROI II area for the actual vehicle speed between 20-40 km/h, and for vehicle with actual speed between 40-60 km/h, the accuracy was high if it is in the ROI III area.