Target-free automated image orientation and camera calibration in close-range photogrammetry

Abstract

Over recent years, a number of computer vision-based algorithms have been developed for automatic target-free 3D scene reconstruction for narrow-baseline configurations of multiple images. However, the performance of these structure-from-motion approaches is invariably disappointing, with the metric integrity being poor and the number of outliers being excessive. This paper presents an approach that overcomes these shortcomings in that it accommodates high-accuracy markerless automatic multi-image orientation and camera calibration within closerange photogrammetric networks that incorporate wide-baseline, highly convergent images. The first stage in the process is feature point detection, fo..