Abstract
Stone flakes are assembled on the same stone core and the position and posture of the adjacent flake surfaces are restored to form a joining material of a stone tool. The operation to assemble stone tool flakes is usually carried out for a population group of materials excavated from a single archaeological ruin, but the group does not always contain all of necessary stone flakes to restore the original stone tool. It is difficult to try assembly for all the excavated stone flakes for the restoration. Due to this, it is hard to determine when to terminate the assembly of stone flakes, or the conditions to terminate the assembly. In addition, the assembly is generally performed manually through a trial-and-error process, which becomes a heavy burden of workers. A method has been proposed for extracting flake surfaces of stone tools and detecting the matching ones to be adjacent by using a computer. With the conventional method, however,it is difficult to detect adjacent flakes for complex flake surfaces, such as a surface for which multiple flakes are merged. Moreover, matching calculation must be implemented for all the target flake surfaces. This operation is very time consuming to search and apply them to a large number of stone tools. In this study, therefore, we propose a faster search method of matching surfaces for complex flake surfaces. In this method, stone tools are measured automatically to extract point clouds on flake surfaces and the feature quantity is calculated. Next, the feature quantity is used to match the flake surfaces and candidates for adjacent surfaces are detected at high speed. Then, appropriate flakes are selected from the candidates and the orientation optimization method is applied to join the flakes. Finally, the flake surfaces are analyzed and two flake surfaces are merged, and we verify that it is possible to automate stone tool material creation from complex flake surfaces.
