Registro tridimensional de faces com fins forenses

Abstract

This study explored the application of photogrammetry to create three- dimensional (3D) models of human faces in specific camera settings. The methodology involved capturing images of mannequins simulating human faces positioned at different distances and angles from the cameras. The data were processed using Agisoft Metashape Standard (version 2.0.0) software to create 3D models of the faces. The results revealed that the configuration with a 60 cm distance between the mannequins and 40 cm between the cameras provided the best results in creating 3D models. In this scenario, it was possible to capture 70 images in 5 different positions, resulting in high-quality 3D models. However, small imperfections were still observed in specific regions, such as the nose. The application of these three-dimensional models was discussed in forensic and clinical contexts, highlighting their potential to improve the comparison of images from various perspectives. However, the study also emphasized that the quality of security cameras could negatively affect the quality of the models. In conclusion, this study demonstrated that photogrammetry can be an effective tool for creating 3D models of human faces in specific camera configurations. Despite challenges related to image quality, these models have the potential to significantly contribute to forensic applications, such as facial recognition, and in the clinical field for detailed patient records. Future research may focus on further refining this technique and testing it on humans to validate its practical applicability