Comparação de malhas triangulares reconstruídas a partir de nuvem de pontos com o uso de métrica da qualidade do elemento

Abstract

Virtual Reality (VR) is an advanced interface that interacts with computer systems where the user can visualize, move, and interact in a three-dimensional environment in real time. This environment is an element of shapes representing the real or abstract world defined by its geometry, color, and lighting. With the creation of three-dimensional models, the user can interact with these systems by visualizing the environment, and the image can be presented on a monitor, in the case of non-immersive systems, or with the aid of glasses or virtual reality helmets to virtual reality immersive systems. Using VR systems to build digital twins of real cities that allow society’s participation and collaboration in decision-making processes is one of the benefits that the advancement of this technology brings to urban planning and monitoring in smart cities. Getting data from the real world is part of the construction of environments for 3D visualization and has tools such as 3D scanners that capture the x, y, and z coordinates of points to files called Point Clouds. To make realistic virtual models, one must first reconstruct the sampled surface in the Point Cloud and then present it with the support of VR-compatible systems and devices. To increase the fidelity in the representation of virtual models of environments, it is necessary to determine the quality of the reconstruction methods and compare them. Therefore, the radius-ratio 𝜌2 metric was used with Ball-Pivoting Algorithm Reconstruction method and Greedy-Projection Triangulation applied to public datasets of objects already analyzed in other related works. Histograms were generated to analyze the frequency distribution of the metric values in order to compare the reconstructed meshes. In the analyzed results, the Ball-Pivoting Algorithm reconstruction method generated better quality meshes compared to the Greedy-Projection Triangulation method.