This thesis proposes a structured and scalable workflow for semantically enriched Smart Point Cloud (SPC) grounded in Heritage Building Information Model (HBIM) ontology. Rather than representing the heritage object with vector-based parametric models, this approach treats the smart point cloud itself as a valid HBIM geometry representation, preserving geometric fidelity while attaching multi-layered semantic information at the patch level. A structured semantic model is defined through a literature-based ontology review, encompassing structural, material, historical, cultural, and conservation-related characteristics. The SPC workflow is implemented and tested on two heritage case studies: the Herdenkingsmonument Kartuizerklooster and the Aula of TU Delft. Each case demonstrates the generality of the method under different geometric and semantic complexities. The semantic annotations are stored externally in structured JSON files, ensuring modularity, version control, and future interoperability. A lightweight web-based viewer was developed using Three.js to support interactive visualization and interpretation, enabling users to explore structure, material, and cultural information directly in the browser. Although full integration with 3D Gaussian Splatting (3DGS) could not be achieved due to current toolchain limitations, the thesis outlines strategies for propagating patch-level semantics to 3DGS centers, as well as segmenting and visualizing per patch with Gaussian Splatting, establishing groundwork for future research in full semantically integrated rendering. Overall, this study contributes a reproducible methodology for documenting, interpreting, and disseminating heritage datasets in a way that aligns with HBIM objectives while minimizing modeling overhead. The data processing and the visualization platform are shared on Github by https://github.com/Zhuoyuee/thesis and https://github.com/Zhuoyuee/spc viewer/tree/main.

This study explores the effectiveness of Virtual Reality (VR) compared to the use of 2D interfaces in interpreting point cloud data, focusing on user perception, interaction and relative measurement accuracy. Visualizing point clouds is often challenging due to the limitations in translating three-dimensional data into two-dimensional screens. VR offers a potential solution to enhance depth perception and deepen user understanding. The research utilizes Omnibase, a platform developed by Geodelta, that integrates various spatial data types, including point clouds, for applications such as municipal boundary measurements.

The study involved participants that are either familiar or unfamiliar with point clouds, to evaluate VR versus Omnibase. Quantitative measurements and qualitative feedback were collected on either platform. Results indicate that while VR provides better depth perception and a more immersive experience, it presents a steeper learning curve, especially for inexperienced users, additionally, it comes with physical side effects. The measurements in Omnibase showed higher consistency, though not necessarily greater accuracy, due to depth misinterpretations.

In addition to the study, the VR testing environment was developed using Potree.