Vehicle Pose Measurement with a Single Image via the Double-Dual-Ellipse Feature of Wheel Rims

The precise measurement of the vehicle pose is a key technology for achieving the intelligent inspection and understanding of vehicle behavior. This paper presents a vehicle pose measurement method utilizing a line-space dual-ellipse representation of wheel rims. First, a dual quadratic curve formulation is established to model wheel-rim projection ellipses, where the common tangent extraction is transformed into a generalized eigenvalue problem, enabling analytical vehicle rotation recovery from a single image. Second, a metric vehicle translation recovery scheme is developed by integrating the vehicle's longitudinal wheelbase constraint. The research conducts comprehensive and systematic experimental verification, including simulation, comparative analysis, vehicle test, and the physical benchmark experiment. The experimental results demonstrate that the proposed method achieves high measurement accuracy and reliability, providing a solution for vehicle pose measurement in practical intelligent vehicle inspection.