Feature-Enhanced Visualization of 3D Point Clouds of Cultural Heritage in Transparent Virtual Reality

Abstract. While virtual reality (VR) has been increasingly adopted for the digital archiving of cultural heritage, most existing systems focus on visual appreciation rather than structural analysis. To address this limitation, this study proposes a transparent VR visualization system designed to enhance the recognition of structural features in 3D scanned point clouds of historic architecture. Using a large-scale point cloud dataset of the main hall of Tamaki Shrine, which is part of the UNESCO World Heritage site “Sacred Sites and Pilgrimage Routes in the Kii Mountain Range,” the data were initially processed using Poisson disk sampling to optimize real-time rendering throughput while preserving geometric fidelity. Structural salience was subsequently quantified through principal component analysis (PCA), employing linearity for sharp-edge extraction and the change of curvature for soft-edge representation. To alleviate the occlusion inherent in opaque point cloud rendering, we implemented a custom alpha-blending shader in Unity. This enabled two types of transparent feature-enhanced visualization: binarization for highlighting sharp edges and opacity gradation for visualizing soft-edge regions. The developed system, operated with a Meta Quest Pro, provides distinct viewing and analysis modes. Experimental results demonstrate that combining feature enhancement with opacity control improves the visibility of internal depth relationships and point-density variations in complex wooden architecture, thereby supporting structural understanding that is difficult to achieve with conventional rendering.