Open AccessImproved point clouds from a heritage artifact depth low-cost acquisition
Author(s) -
Pedro Raimundo,
Karl Apaza-Agüero
Publication year - 2020
Publication title -
revista brasileira de computação aplicada
Language(s) - English
Resource type - Journals
ISSN - 2176-6649
DOI - 10.5335/rbca.v12i1.10019
Subject(s) - point cloud , artifact (error) , computer science , computer vision , noise (video) , object (grammar) , artificial intelligence , point (geometry) , data acquisition , computer graphics (images) , image (mathematics) , mathematics , geometry , operating system
Poor data acquisition from low-cost cameras, such as low-resolution depth maps or high level of noise or point clouds generated with insufficient information from an object, limits the use of such cameras for heritage artifacts 3D reconstruction. This work proposes to improve this depth low-cost acquisition by using a new approach based on the Super-Resolution technique. The proposed approach has been applied to several artifacts of the Federal University of Bahia Museum of Archaeology and Ethnology (MAE/UFBA). As shown in the results, our approach improved the quality of point clouds generated from tested heritage artifacts. Results indicate that whenever artifact geometry is gained via our method there is actual reconstruction of detail or accuracy improvements, whereas a reduction in number of points of the clouds, if any, would indicate the removal of inconsistencies or noise from the input data without loss of detail.