Car speed estimation based on cross-ratio using video data of car-mounted camera (black box)

Area of Science:

• Computer Vision
• Automotive Engineering
• Transportation Safety

Background:

• Estimating vehicle speed from dashcam footage is crucial for traffic analysis and accident reconstruction.
• Current methods often rely on GPS data or known environmental markers, limiting their applicability in diverse scenarios.

Purpose of the Study:

• To develop and validate methods for estimating car speeds using only car-mounted camera (black box) footage.
• To enable direct velocity estimation without requiring specific physical locations or GPS data.

Main Methods:

• Collected and classified 96 cases of black box footages under various travel circumstances.
• Conducted case studies for speed estimation of both the camera-mounted car and other vehicles.
• Developed a method for estimating speeds of vehicles on curvilinear paths.
• Utilized cross-ratio calculations for speed estimation.

Main Results:

• Demonstrated the feasibility of estimating car speeds directly from black box footage.
• Validated the cross-ratio method against traditional footage analysis and GPS data for camera-mounted cars.
• Showcased applicability for stationary and moving camera-mounted cars, as well as other vehicles.

Conclusions:

• The proposed methods offer a practical solution for estimating vehicle speeds from readily available black box recordings.
• This approach enhances the utility of dashcam footage for traffic monitoring and analysis without external data sources.