Stone Tool Identification Method Based on Measured Points by RGB-D Camera and Points of Stone Tool Database

Abstract

石器データベースと RGB-D

A method to improve the accuracy of stone tools identification with D2 distribution

“…The database used in the matching process is constructed in the same way as presented in the previous study. 1 The database constructed in the previous study, is composed of silhouette images and 3D point clouds of stone tools, which are obtained when stone tools are looked down from directly above at all postures where they are stable if placed on a flat surface. By building a database for all stable postures of each stone tool, identification is possible no matter what direction the tool is placed when measured with the camera.…”

“…The previous study 1 proposed an identification method that realizes matching using stone tools silhouettes 2 and the ICP algorithm. However, it is difficult to identify stone tools if they are thick because the matching method is based on two-dimensional approach.…”

“…However, when using 3D information, the identification speed becomes a problem. For example, the previous study 1 reported that 3D identification between stone tool pairs took several seconds, which was time-consuming when evaluating multiple pairs. In contrast, Chida et al 3 used D2 distribution, which has 3D features, as a preprocessing step for stone tools matching to narrow down the candidates and obtain a speedup effect.…”