光流测距全向相机的标定与三维重构

Calibration and 3D reconstruction with omnidirectional ranging-by-optic-flow camera

针对基于主动式传感器的无人驾驶飞机导航系统在实际应用中存在的体积庞大、成本高等问题,介绍一种由传统透视相机和具有等距性的反射镜面组成的相机镜面系统.镜面的等距性使得空间点到相机光轴的轴心距与图像光流值之间存在一种简单的对应关系,使用该相机镜面系统可以直接对场景进行测距.通过仿真实验发现,系统测距精度主要受镜面与相机的角度安装误差的影响.提出一种新的参数标定方法来改善测距精度.结果表明:将相机应用于实际场景,可以较好地实现环境的三维重构,证明所提出的标定与重构方法的有效性.

There are limitations when applying active sensing in designing guidance systems for UAVs,which can be bulky and expensive.An omnidirectional ranging-by-optic-flow camera was introduced,which was composed of a traditional perspective camera and a reflective mirror with certain imaging characteristic.The imaging characteristic helped establish a simple relationship between the distance from space point to the optical axis of camera and the value of optic flow.Thus,3Dreconstruction of test scene could be realized directly.As a result,the ranging accuracy of the system is mainly influenced by the angular error of installation through simulation experiments.Therefore,a new parameters calibration method was proposed to assist in improving ranging accuracy.Results show that 3Dreconstruction can be achieved by exploiting the designed camera to the stereo scene.The experimental results demonstrate the effectiveness of the proposed calibration and reconstruction method.