Traffic-Oriented Three-Dimensional Vehicle Reconstruction Using Fixed Roadside Monocular Camera Sensors

TL;DR

A new framework uses roadside cameras to accurately reconstruct 3D vehicle shapes in real-world traffic, improving efficiency and accuracy.

Contribution

A traffic-oriented framework that combines semantic and non-semantic features for efficient 3D vehicle reconstruction from monocular cameras.

Findings

The method reduces mean reprojection error by 13.6% compared to conventional SfM methods.

Reconstruction time is shortened by 43.9% using the proposed framework.

Abstract

What are the main findings?
A traffic-oriented framework enables reliable 3D vehicle reconstruction using fixed roadside monocular camera sensors under real-world traffic conditions.Joint use of semantic and non-semantic feature points significantly improves reconstruction accuracy and efficiency compared with conventional incremental SfM methods.

A traffic-oriented framework enables reliable 3D vehicle reconstruction using fixed roadside monocular camera sensors under real-world traffic conditions.

Joint use of semantic and non-semantic feature points significantly improves reconstruction accuracy and efficiency compared with conventional incremental SfM methods.

What are the implications of the main findings?
Existing roadside camera infrastructures can be leveraged to enhance three-dimensional traffic perception without additional sensing hardware.The proposed approach supports scalable deployment of camera-based sensing systems for intelligent transportation applications such as traffic monitoring and analysis.

Existing roadside camera infrastructures can be leveraged to enhance three-dimensional traffic perception without additional sensing hardware.

The proposed approach supports scalable deployment of camera-based sensing systems for intelligent transportation applications such as traffic monitoring and analysis.

Fixed roadside monocular cameras are widely used as low-cost sensing devices in intelligent transportation systems; however, extracting reliable three-dimensional (3D) information from such sensors remains challenging due to limited baselines, long observation distances, and moving vehicles. This paper presents a traffic-oriented 3D vehicle reconstruction framework based on monocular image sequences captured by fixed roadside camera sensors. Semantic and non-semantic vehicle feature points are jointly exploited to balance structural consistency and surface completeness, and a feature-map-consistency-based optimization strategy is introduced to refine feature point localization and reduce reprojection errors. In addition, an optimized incremental Structure-from-Motion (SfM) pipeline incorporating traffic-aware initialization, keyframe selection, and local bundle adjustment is developed to improve reconstruction efficiency. Experiments on real-world traffic surveillance videos show that the proposed method reduces the mean reprojection error by 13.6% and shortens reconstruction time by 43.9% compared with widely used incremental SfM systems.

Figures