실내 바닥면 포인트 클라우드를 이용한 AI 기반 3D 공간구조 추정 기술

AI-based 3D Spatial Structure Estimation Using Indoor Floor Point Cloud

초록

In this paper, we construct a virtual space composed of point clouds by scanning real-world indoor spaces using RGB-D cameras. Point clouds have distributed and noisy properties, making it difficult and a very challenging task to reconstruct a virtual space similar to the real world (using point clouds). We propose an efficient method for estimating and modeling the structure of 3D space based on CNN model. Based on the 3D reconstruction algorithm, an indoor point cloud is created through RGB-D images and floor data is extracted. Floor point cloud data is projected and converted into 2D binary images, and corners of floor is detected using CNN model. Based on this, the 3D spatial structure of the indoor space is estimated and modeled. In the future, this study will contribute to efficiently establishing a metaverse space similar to the real world.

키워드

3D reconstruction; Spatial Structure Estimation; Digital Twin; Coner Detection; Metaverse; 3D 재구성; 공간구조 추정; 디지털 트윈; 코너 탐지; 메타버스
제목
실내 바닥면 포인트 클라우드를 이용한 AI 기반 3D 공간구조 추정 기술
제목 (타언어)
AI-based 3D Spatial Structure Estimation Using Indoor Floor Point Cloud
저자
유재석; 유태경; 서상현
DOI
10.9728/dcs.2022.23.6.1013
발행일
2022-06
저널명
디지털콘텐츠학회논문지
권
23
호
6
페이지
1013 ~ 1020