The impact of camera calibration on UAV-based photogrammetric products using a non-metric camera

By Persa Koutsouradi

Abstract A range of consumer-grade digital cameras, which are also known as non-metric cameras, have become the most widely used camera sensors in UAV-based photogrammetry. Nonetheless, a major limitation of non-metric cameras is the camera lens distortion that non-metric cameras often encounter. Camera lens distortion can cause geometric instability in the captured UAV images, which in turn can negatively affect the photogrammetric products’ quality. Geometric stability refers to the ability of a camera to keep the images’ geometry the same over time. A common way to ensure geometric stability in the derived UAV images is to geometrically calibrate the non-metric camera. In practice, there are two main approaches towards calibrating a non-metric camera: independent of the photogrammetric process (pre-calibration) or inside the photogrammetric process (self-calibration). The pre-calibration can be further divided into lab and in-situ (in the field). Despite the numerous studies investigating the impact of camera calibration on the geometric accuracy of photogrammetric products, it remains unclear whether the in-situ pre-calibration provides higher image geometric stability compared to self-calibration when the flight altitude alters. In addition, there is a lack of research examining the impact of in-situ pre-calibration and self-calibration on different landcover surfaces. Based on the identified research gap, the objective of this research is to perform a comparative analysis of the impact between in-situ pre-calibration and self-calibration on the geometry of UAV photogrammetric products, using a non-metric camera. Regarding the land monitoring scenarios, this research was conducted at three different flight altitudes (60m, 90m, and 120m) and over an agricultural, forest, and artificial landcover surface.

The findings suggest the following: (1) the in-situ pre-calibration shows higher geometric stability of the interior orientation parameters compared to self-calibration at 60m, 90m and 120m altitude, (2) the in-situ pre-calibration has a more positive impact than self-calibration on both the horizontal and vertical reconstruction of 2D and 3D agricultural and artificial surfaces at low altitudes (60m), (3) the in-situ pre-calibration has a more positive impact than self-calibration on the horizontal reconstruction of 2D and 3D forest surface at high altitudes (120m) and (4) the in-situ pre-calibration has a more positive impact than self-calibration on the vertical reconstruction of 2D forest surface at low altitudes (60m).

This research has provided insights into the impact of in-situ pre-calibration on reducing the geometric instability in UAV photogrammetric products. It would be worthwhile for future research to compare the impact of lab pre-calibration and in-situ pre-calibration on the UAV-based photogrammetric products on similar land monitoring scenarios.

Keywords: non-metric camera; UAV-based photogrammetry; in-situ pre-calibration; self-calibration; land monitoring scenarios