A Comparitive Study of 3D Digital Reconstruction for Indoor Static Scenes Based on RGB D Data

Abstract

3D digital reconstruction of indoor static scenes is a key requirement in many areasˈsuch as indoor navigation, augmented reality, heritage restoration and computer vision, etc. The popularity of current low-cost RGB-D sensors has opened new lines of research in indoor 3D reconstruction. We focus on an experimental review of state-of-the-art 3D digital reconstruction methods using RGB-D images. Specifically, we propose the research goals and core issues in this field, summarize the acquisition methods of RGB-D data and some popular RGB-D datasets, and divide the 3D reconstruction techniques based on RGB-D into geometry-based and primitive-based categories, in addition, to better compare current researches in the area, some existing methods are experimentally evaluated on the same test datasets. Finally, respect to the open research problems in this field are discussed.