A photogrammetric network and object field design for efficient self-calibration of non-metric digital cameras

Recent increase in the number of pixels of a non-metric digital camera encourages to use it for close-range photogrammetry such as modeling cultural asset and buildings. However, these cameras have to be calibrated for close-range photogrammetry application. For self-calibration, an appropriate photogrammetric network and object field should be designed. In this paper, we studied the effect on self-calibration accuracy changes according to the change of the number of ground control points, dimensions of the ground control points, and the combination of images. We concluded that self-calibration with three photos including a vertical photo can give the stable accuracy of interior orientation parameters and 10 ground control points on a plane can give high accuracy for object reconstruction.