배관 인식을 위한 마커리스 증강현실 시스템 기반의 깊이에 대한 연구

초록

In this study, a depth based Pipeline’s Recognition Algorithm was proposed for the future markerless AR system. Even if the pipeline system onboard ship is very complex, according to theirs painting color and diameter, they are also well ranged. In this algorithm, a RGB-D scanning data of a lab-test pipeline system was firstly captured. After de-noising, down sampling and detection, pipes are respectively segmented by different diameter and different color. And then by registered with pipe’s design point cloud which converted from CAD model system to figure it out which part of these segmented pipes belonging to. Lastly, these registered pipelines will art as the recognition component for markerless AR system.