抄録
Traditional archaeological excavations often record only limited details of artifacts and sites, making postexcavation analysis difficult. A novel approach, consistent 3D excavation, records every detail of the excavation process by scanning both artifacts and the excavation site to create 3D models. However, the current method of registering the artifact and site models is entirely manual operation, requiring significant labor and time, and the result depends heavily on the operator’s skill. This paper proposes a method to automate the registration of 3D point clouds from an archaeological trench investigation site with individual artifacts (stones) in a virtual space. The method integrates histogram-based color segmentation and a region-growing algorithm to extract the topside of stones from the trench point cloud. The segmented stone points are then registered with the complete stone point cloud, using initial alignments to refine the Iterative Closest Point (ICP) results, creating a virtual representation of the ruin. To evaluate the effectiveness of the proposed method, experiments were conducted on two different trench datasets. The results show that the method achieves high segmentation and registration accuracy, while significantly reducing manual effort and improving efficiency.
A Study on Automatic Registration of 3D Point Clouds Obtained from Archaeological Trench Investigations
Fullsize Image
