Virtual 3D Reconstruction of Archaeological Pottery Using Coarse Registration

The 3D reconstruction of objects has not only improved visualisation of digitised objects, it has helped researchers to actively carry out archaeological pottery. Reconstructing pottery is significant in archaeology but is challenging task among practitioners. For one, excavated potteries are hardly complete to provide exhaustive and useful information, hence archaeologists attempt to reconstruct them with available tools and methods. It is also challenging to apply existing reconstruction approaches in archaeological documentation. This limitation makes it difficult to carry out studies within a reasonable time. Hence, interest has shifted to developing new ways of reconstructing archaeological artefacts with new techniques and algorithms. Therefore, this study focuses on providing interventions that will ease the challenges encountered in reconstructing archaeological pottery. It applies a data acquisition approach that uses a 3D laser scanner to acquire point cloud data that clearly show the geometric and radiometric properties of the object’s surface. The acquired data is processed to remove noise and outliers before undergoing a coarse-to-fine registration strategy which involves detecting and extracting keypoints from the point clouds and estimating descriptions with them. Additionally, correspondences are estimated between point pairs, leading to a pairwise and global registration of the acquired point clouds. The peculiarity of the approach of this thesis is in its flexibility due to the peculiar nature of the data acquired. This improves the efficiency, robustness and accuracy of the approach. The approach and findings show that the use of real 3D dataset can attain good results when used with right tools. High resolution lenses and accurate calibration help to give accurate results. While the registration accuracy attained in the study lies between 0.08 and 0.14 mean squared error for the data used, further studies will validate this result. The results obtained are nonetheless useful for further studies in 3D pottery reassembly.