Análisis de la precisión entre modelos 3d internos generados con nubes de puntos de alta densidad

Abstract

The development and expansion of scanning technologies have led to the acquisition of increasingly dense point clouds, enabling the creation of 3D models with a remarkable level of accuracy. However, the actual quality of these models depends directly on how faithfully the retrieved data represents the true characteristics of the world being studied. Therefore, it is essential to rigorously evaluate the accuracy of the acquired models and verify their fidelity to the construction of the physical space they are intended to describe. This monograph aims to analyze the accuracy and reliability of internal 3D models generated from high-density point clouds. The goal is to determine their technical validity and relevance in operational processes and spatial documentation. To this end, widely recognized statistical methods for geometric validation are employed, such as the root mean square error (RMSE), the paired t-test, Pearson's correlation coefficient, and Bland-Altman analysis. These tools allow for the quantification of errors, the identification of potential biases, and the evaluation of the level of agreement between the compared models. The methodology presented begins with the generation of 3D models from internal scanning, followed by comparison against spatial references and the detection of any discrepancies with the real-world environment. This approach allows not only for the verification of the geometric quality of the data but also provides the capacity to establish acceptance criteria for subsequent use. However, the process presents some challenges, including the availability of reliable reference points, variations in capture conditions, and the inherent technical nature of the equipment used. Ultimately, the study aims to determine the degree of accuracy of the evaluated 3D models and statistically verify their correspondence with the physical space. In doing so, the study seeks to provide technical evidence supporting the use of high-density point clouds in spatial documentation processes, ensuring that the generated models meet appropriate standards of quality and geometric fidelity