Camera Self-Calibration in UAS Photogrammetry – When is it Viable

Clive Fraser

Prof. Clive S Fraser University of Melbourne

22F

Over the past few years there has been a significant increase in the adoption of aerial photogrammetry from UAS/UAV/drone platforms as a fully automatic 3D object reconstruction and mapping tool, across application domains as varied as local-area mapping and deformation monitoring in the mining industry, through to construction monitoring, heritage recording and the modelling of heritage buildings and artefacts. In many cases – perhaps even the majority – the photogrammetric surveys are being performed by non-specialist users who treat the photogrammetric processing as a black-box operation. This is all well and good when the metric quality of the sensor system and the geometric strength of network configuration yield an accurate and reliable result, but shortcomings in camera calibration can cause significant accuracy underperformance in UAS photogrammetry. There is widespread use of off-the-shelf, consumer-grade grade cameras with unstable interior orientation and consequently unstable metric calibration. Self-calibration can greatly enhance the integrity of photogrammetric measurement, but the carrying of additional parameters in a bundle adjustment can be a fraught exercise in cases where the network geometry is not conducive to the accurate recovery of camera parameters, especially interior orientation parameters. It has long been recognised within the photogrammetric community that aerial blocks of nadir imagery offer the least suitable network geometry for self-calibration, yet this is the very configuration most widely used in UAS projects. This presentation discusses camera self-calibration for UAS photogrammetry, and considers in what circumstances it will be viable. Non-viability here implies that adoption of self-calibration is likely to lead to significant positional biases and shape distortion in the generated 3D point cloud in object space. Practical steps which can be taken to optimize scene-independent camera self-calibration are also discussed. These steps aid in avoiding accuracy degradation that can arise through projective compensation. The subsequent introduction of systematic positioning errors in 3D object reconstruction, principally biases, can then also be avoided.

Related Divisions: UAS Division, Photogrammetric Applications Division

10:30 Camera Self-Calibration in UAS Photogrammetry – When is it Viable, Clive Fraser

January 30 @ 10:30
10:30 — 10:45 (15′)

Granite ABC

Clive Fraser

Add to Google Calendar