Spherical images for the survey of indoor environments

Abstract

In recent years there has been a considerable increase in the use of spherical images for the generation of three-dimensional models.
Nowadays there are many software (commercial or open source) dedicated to the processing of images, but only some of them could
process this data; in particular, the common software based on computer vision algorithms are not able to calculate the interior
orientation parameters of these systems. This article attempts to analyze the potential of spherical images processing for the
reconstruction of three-dimensional objects through Structure from Motion (SFM) algorithms, comparing the results with those
obtained from the use of the individual plane images, and to evaluate the possibility to calculate with the same software the
calibration parameters of the single cameras. Some tests were conducted in indoor environment through the use of the NCTech
iSTAR Fusion panoramic camera. The reference model for the analysis on the achieved accuracy is a LiDAR point cloud, with which
the three-dimensional products will be compared. In addition, an evaluation on the interior orientation parameters calculated from
different software will be made.