Research Article
FSE-MV: Compressed Domain Video Information Assisted Hybrid Real-Time Vehicle Speed Estimation
@INPROCEEDINGS{10.1007/978-3-030-94763-7_8, author={Yangjie Cao and Qi Wu and Bo Zhang and Zhi Liu and Junfeng Li}, title={FSE-MV: Compressed Domain Video Information Assisted Hybrid Real-Time Vehicle Speed Estimation}, proceedings={Mobile Networks and Management. 11th EAI International Conference, MONAMI 2021, Virtual Event, October 27-29, 2021, Proceedings}, proceedings_a={MONAMI}, year={2022}, month={1}, keywords={Speed estimation ITS Feature matching Compressed domain}, doi={10.1007/978-3-030-94763-7_8} }- Yangjie Cao
Qi Wu
Bo Zhang
Zhi Liu
Junfeng Li
Year: 2022
FSE-MV: Compressed Domain Video Information Assisted Hybrid Real-Time Vehicle Speed Estimation
MONAMI
Springer
DOI: 10.1007/978-3-030-94763-7_8
Abstract
Vehicular speed estimation is a vital component in intelligent transportation systems. With the recent development of smart cameras and computer vision technologies, video-based vehicle speed estimations have been widely studied. However, facing the huge volume of pixel-domain information, conventional methods are computationally intensive, and often fail to deliver estimation results in real-time. In this paper, we target the video-based real-time vehicle speed estimation problem. For data volume reduction, we utilize the compressed domain video information and propose a hybrid real-time vehicle speed estimation method termed FSE-MV. FSE-MV first segments vehicles using motion vector (MV) information in the compressed domain. The pixel information of the segmented vehicles is then retrieved through decoding. Feature points of each vehicle are extracted for multi-object matching and pixel domain displacement calculation. The speed of the target vehicle is finally calculated through spatial coordinate transformation. Experiments over the public dataset demonstrate that FSE-MV is able to process 1080p traffic video data in real-time ((\thicksim )30 frames per second) with a high estimation accuracy ((\thicksim )93.09%).
