Calibration of a vehicle camera system with divergent fields-of-view in an urban environment

Multi camera systems are used in modern cars for environment perception. To combine information extracted from different cameras, their relative pose has to be known. Therefore, a multi camera system can be calibrated. In this contribution, a calibration approach using the structure-from-motion (SfM) method is described. From images of urban structures, 3d points are calculated using SfM. These 3d points are used as reference to estimate the pose of the vehicle cameras. Test images are taken in a car park with a DSLR camera, together with two action cameras mounted in a test vehicle recording the car park. More than 26,000 3d points representing the car park can be obtained. With these points, the poses of the vehicle cameras are estimated successfully. The distance between the vehicle cameras shows the highest deviation from laser reference measurements for images showing a low point density area of the car park.     «

Multi camera systems are used in modern cars for environment perception. To combine information extracted from different cameras, their relative pose has to be known. Therefore, a multi camera system can be calibrated. In this contribution, a calibration approach using the structure-from-motion (SfM) method is described. From images of urban structures, 3d points are calculated using SfM. These 3d points are used as reference to estimate the pose of the vehicle cameras. Test images are taken in...     »