The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLIII-B2-2020
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B2-2020, 487–494, 2020
https://doi.org/10.5194/isprs-archives-XLIII-B2-2020-487-2020
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIII-B2-2020, 487–494, 2020
https://doi.org/10.5194/isprs-archives-XLIII-B2-2020-487-2020

  12 Aug 2020

12 Aug 2020

RESECTION AND MONTE CARLO COVARIANCE FROM VANISHING POINTS FOR IMAGES OF UNKNOWN ORIGIN

R. Settergren R. Settergren
  • BAE Systems, Geospatial eXploitation Products™ (GXP®), 10920 Technology Drive, San Diego CA, USA

Keywords: photogrammetry, vanishing points, parallel lines, resection, 3D scene reconstruction

Abstract. Photogrammetric analysis requires camera metadata (position, attitude, interior orientation, etc.), which is not available for all images. Modern commercial solutions for 3D reconstruction from images typically assume large amounts of purposefully-collected, highly-overlapping imagery. In this article, a system is demonstrated for recovering a 3D scene from images of unknown origin, by marking ground space axes, resecting a camera consistent with the markings, and then using the solved camera to collect 3D measurements. The capability works with close-range (vanishing points) and long-range (parallel axes) imagery. Monte Carlo analysis is used to propagate measurement uncertainty of the vanishing lines to uncertainty of exterior and interior camera parameters, which can then be used to quantify uncertainty of measurements in the 3D scene.