Reconstructing building surface models with enhanced geometric feature constraints

This research introduces a new reconstruction algorithm of building surface models that help in improving the efficiency and reliability of the models by incorporating geometrical feature attributes. However, the traditional reconstruction techniques fail when geometries are complex or have irregular patterns resulting in a partially reconstructed model or a model with errors. The problems mentioned above are resolved by the proposed algorithm that utilizes geometric feature constraints which allow to guide the reconstruction process and recover important architectural features. Data for the methodology are captured using 3D laser scanners and photogrammetry; the features are then extracted from edges, and surfaces using various techniques which include surface fitting; the constraints are formulated to maintain geometric compatibility. The implementation of the algorithm is done in Python, and the results are tested on various objects of different typology, such as historical and architectural, business, and various tall structures. The findings show enhanced levels of the surface deviation, edge precision, planar surfaces uniformity, and time consumption that is far from the conventional methods. Some of the findings are Mean surface deviation which is equal to 2.5mm for the monument and 95% for the edge detection feature of modern buildings and lastly, 95% for the planar surface consistencies. By focusing on the practicality of geometric features constraints in the reconstruction of buildings and stressing on its usefulness, the research offers its contribution to provide exact and credible models for uses in structures’ architecture, urban designing, computer illustration, and conservation of historical structures. Possible future works include the application of supervised/unsupervised machine learning to improve feature extraction and the use of real-time data collection and processing to update the models constantly. It is in this regard that this study offers a good foundation for establishing improved reconstruction algorithms that would serve the developing necessities of the industry.