The Influence of the Point Cloud Comparison Methods on the Verification of Point Clouds Using the Batavia Reconstruction as a Case Study

Abstract

During the excavation of the Batavia shipwreck in the 1970s part of the wooden hull and additional artefacts were excavated and conserved. In the course of the excavation over 3500 underwater images were captured to document the expedition. These images have allowed for the digital 3D reconstruction of the Batavia as it was found on the sea floor. The physical shipwreck timbers were reconstructed and are now located in the Fremantle Shipwrecks Museum. In 2014, a Terrestrial Laser Scanner (Leica C10) was utilised to capture the shipwreck as displayed in the museum. This resulted in two different point clouds of the vessel: a) based on the images captured underwater and, b) based on laser scan data captured in the museum. The case study will focus on two different sections of the vessel with different complexity, and has two aims. Firstly, the goal is to compare the 3D reconstructions from the underwater archival photography with the laser scans and 3D reconstruction of the Batavia hull currently on display to verify if the Batavia as displayed today has the same configuration as the original wreck. The results show that while the difference between both datasets is often minimal, some differences are visible because the excavation took place over a long period. Secondly, the paper will evaluate the influence of different methods for point cloud comparison. The methods are namely point-to-point and point-to-model and are used to quantify the differences of the shipwreck's point clouds. Even though the same input point clouds were utilised, the paper will show that the results of the comparison were influenced by the point-cloud-comparison method used.