Voxel Based Web Visualisation for 3D Cadastre in Resource Constrained Urban Environments

Abstract

Rapid urban verticalization has made conventional two-dimensional cadastral systems inadequate for representing overlapping property rights, vertical restrictions, and building-volume relationships in dense urban environments. This study aims to develop and evaluate a lightweight web-based 3D cadastral visualisation framework that can support accessible land administration in resource-constrained settings. Using Sumur Bandung District, Bandung, Indonesia, as a case study, the research processed 6,762 building footprints covering 3.39 km² through a hybrid Python-JavaScript voxelisation pipeline. The proposed method converts 2D cadastral vector footprints into semantic volumetric units at a 0.5-metre spatial resolution using a server-side open-source stack based on Flask, GeoPandas, and Shapely, while client-side visualisation is implemented through Three.js and WebGL. System performance was assessed by comparing the voxel-based output with a baseline high-polygon surface mesh generated from the same vector data in QGIS 3.36 and exported as an uncompressed Wavefront OBJ file containing approximately 1.24 million triangular faces. The results show that the proposed framework reduced the data payload from 45.5 MB to 18.2 MB, equivalent to a 60% reduction, while maintaining interactive rendering at approximately 55 FPS on a low-specification device equipped with an Intel i5 12th-generation processor and Intel UHD GPU. The system achieved a Time-to-Interactive of 3.2 seconds, improving web accessibility compared with previously reported web-based 3D cadastral solutions exceeding 10 seconds of initial loading time. Geometric accuracy assessment indicated a mean boundary aliasing error of 0.21 m and a maximum error of 0.35 m, remaining within the 0.50 m positional tolerance for Class 3 cadastral data under Indonesian BPN technical guidelines. In addition, the framework integrates static rule-based height-compliance validation against RDTR and KKOP regulations. Validation using 522 manually verified buildings produced 100% precision and 91.7% recall. These findings demonstrate that voxelisation offers a practical, scalable, and open-source approach for modernizing 3D land administration in developing countries.