中文摘要：

由于车轮中心检测是实现车辆轮距、轮距差及车轮静立半径准确测量的关键，提出了一种基于立体视觉的车轮中心高精度检测方法．首先，根据先验知识提取车轮轮毂的椭圆特征，并进行椭圆拟合以获取其椭圆方程．然后，引入非相切约束条件以提高椭圆匹配的精度．最后，采用具有低时间复杂度和高测量精度的空间圆投影方法重建车轮中心的三维坐标．仿真实验表明，相比于椭圆中心重构法和平面约束优化法，所提方法能够更加准确地获取空间圆圆心的三维坐标．此外，通过对3款车型的轮距、轮距差及车轮静立半径的测量实验证明了所提方法对车轮中心检测的有效性．

英文摘要：

As the location of the wheel center is the key to accurately measuring the wheelbase, the wheelbase difference and the wheel static radius, a high-precision wheel center detection method based on stereo vision is proposed. First, according to the prior information, the contour of the wheel hub is extracted and fitted as an ellipse curve, and the ellipse fitting equation can be obtained. Then, a new un-tangent constraint is adopted to improve the ellipse matching precision. Finally, the 3D coordinates of the wheel center can be reconstructed by the spatial circle projection algorithm with low time complexity and high measurement accuracy. Simulation experiments verify that compared with the ellipse center reconstruction algorithm and the planar constraint optimization algorithm, the proposed method can acquire the 3D coordinates of the spatial circle more exactly. Furthermore, the measurements of the wheelbase, the wheelbase difference and the wheel static radius for three types of vehicles demonstrate the effectiveness of the proposed method for wheel center detection.