A novel post-calibration method for digital cameras using image linear features (English)

With the advent of new digital cameras, especially low-weight non-metric ones mounted on unmanned aerial vehicles (UAVs), camera calibration models should be reinforced in accordance with their complex multiple-distortion nature. In this article, a new calibration method is proposed and evaluated, which is highly capable of modelling complex image distortions and needs no additional ground information. This method applies the straight line features to detect and model image distortion via the self-calibration process. This method is then used to upgrade the 3D positional information of existing digital maps in a post-correction scheme. The proposed method was implemented on the metric UltraCam digital images as well as the UAV-based images taken by a non-metric camera. The obtained results were extremely promising because 56% and 58% of planimetric and vertical accuracy improvements were achieved, respectively, in the case of UAV-based images. Similar results were also obtained for the UltraCam project. This level of accuracy can only be achieved in traditional photogrammetric methods with the aid of five times more ground control points (GCPs). The proposed method is extremely effective in terms of the reduction of costs and time consumption due to the decrease in ground operations. This advantage is of great interest, especially in the case of UAV-based operations, where the primary objective is cost reduction. Furthermore, in the case of improving previously generated maps, where the utilization of new ground points is not possible at all, this method is the only applicable approach since it is entirely executed in the image space.