Weighted Graph-Based Assessment of Visual Quality in Triangular Mesh Models
Résumé
This paper focuses on evaluating visual quality in triangular mesh models. We introduce a novel approach using weighted graphs to address this challenge, motivated by the increasing necessity for accurate quality assessment in fields such as computer graphics and 3D modeling. The methodology is initiated by generating saliency maps for each distorted mesh model. Then, we build a network representation where mesh vertices serve as nodes and mesh edges as graph edges. Vertex weights are determined based on salience values, and a diverse set of topological properties are extracted and statistically measured to generate a signature vector. The quality score prediction is conducted through rigorous evaluation of three regression algorithms. Experiments cover four publicly available databases tailored for mesh model quality assessment, showcasing the superior performance of the proposed approach with significant correlations to subjective evaluations.
Origine | Fichiers produits par l'(les) auteur(s) |
---|