|Ph.D Student||Telem Gil|
|Subject||Photogrammetric Models for Underwater Mapping|
|Department||Department of Civil and Environmental Engineering||Supervisor||Professor Sagi Filin|
|Full Thesis text|
Introducing advanced, accurate mapping of complicated scenes makes photogrammetry valuable for underwater mapping. Utilizing photogrammetry benefits a wide variety of applications including seabed mapping, archeological surveys, marine biology, and unmanned underwater vehicle navigation, to name only a few. With current advances in imaging technology, where affordable digital cameras have high quality, using photogrammetry for mapping underwater becomes even more relevant.
Nonetheless, development of appropriate underwater photogrammetric models poses nontrivial modeling problems. Refraction effect, the extension of the imaging system into a unit composed of both camera and a protecting housing device, the lack of fixed reference frames, and imaging from close ranges and in various scales impose significant constraints to the traditional photogrammetric mapping. This thesis studies the effect that underwater environment has on the photogrammetric process and analyzes its theoretical aspects. It proposes a model for describing the geometric distortions and for estimating the additional parameters involved. The proposed model accounts not only for the multimedia effect, but also for inaccuracies related to the setting of the camera and housing device. It targets then the two fundamental photogrammetric principles: collinearity and coplanarity. Due to refraction, both collinearity and coplanarity no longer hold in underwater environments, necessitating development of appropriate mathematical models and theories for the application of photogrammetric concepts in the underwater environment. The thesis shows that only a small number of additional parameters is needed to model both elements and to preserve the collinearity relation, and proposes a closed form model for describing it. The thesis then proposes a model which fulfills the coplanarity condition and thereby requires no knowledge of object space data for computing relative orientation. Aspects of the epipolar geometry are then analyzed upon which an analytical expression is developed to handle the refraction effect. The theoretical analysis is then broadened for mapping of linear objects onto the image. Finally a new representation is derived, which up to a fraction of a pixel's level of accuracy, enables solving the pose parameters in a linear manner.
Results show that the proposed models are robust and facilitate accuracy up to the level of equivalent close-range photogrammetric practices. They also show that no unique setup is needed for estimating the orientation parameters and that the estimation is insensitive to noise or first approximations. The high levels of accuracy, which are obtained, show that the proposed model manages to translate theory into accurate measurement and documentation practices.