Perceptual Crack Detection for Rendered 3D Textured Meshes

TL;DR

This work tackles perceptual crack artifacts in rendered 3D textured meshes by introducing Perceptual Crack Detection (PCD), an HVS-inspired method that derives a crack-likelihood map from distorted versus reference frames using contrast normalization and Laplacian modulation. The approach includes a simple, principled integration framework that re-weights pixel-quality maps from existing IQA models with a crack-aware weight $w_i$ to obtain a final score $Q$, enhancing perceptual quality prediction. Extensive experiments on the Nehmé et al. and TSMD datasets show consistent improvements in SRCC and PLCC across six base QA metrics, with notable standalone performance of CAS, and real-time capable runtime on typical video frames. Overall, PCD enables accurate crack localization and significantly boosts 3D mesh QA by focusing attention on crack-affected regions, facilitating more reliable quality assurance and optimization across the mesh processing pipeline.

Abstract

Recent years have witnessed many advancements in the applications of 3D textured meshes. As the demand continues to rise, evaluating the perceptual quality of this new type of media content becomes crucial for quality assurance and optimization purposes. Different from traditional image quality assessment, crack is an annoying artifact specific to rendered 3D meshes that severely affects their perceptual quality. In this work, we make one of the first attempts to propose a novel Perceptual Crack Detection (PCD) method for detecting and localizing crack artifacts in rendered meshes. Specifically, motivated by the characteristics of the human visual system (HVS), we adopt contrast and Laplacian measurement modules to characterize crack artifacts and differentiate them from other undesired artifacts. Extensive experiments on large-scale public datasets of 3D textured meshes demonstrate effectiveness and efficiency of the proposed PCD method in correct localization and detection of crack artifacts. %Specifically, We propose a full-reference crack artifact localization method that operates on a pair of input snapshots of distorted and reference 3D objects to generate a final crack map. Moreover, to quantify the performance of the proposed detection method and validate its effectiveness, we propose a simple yet effective weighting mechanism to incorporate the resulting crack map into classical quality assessment (QA) models, which creates significant performance improvement in predicting the perceptual image quality when tested on public datasets of static 3D textured meshes. A software release of the proposed method is publicly available at: https://github.com/arshafiee/crack-detection-VVM

Figures (4)

• Figure 1: Sample snapshots of a pair of reference (left) and distorted (right) 3D textured meshes. Crack artifacts are present all over the distorted object. Two local windows are cropped and enlarged for better visualization.
• Figure 2: PCD framework, where TAD represents the truncated absolute difference. The sample reference and distorted frames are obtained from Nehmé et al. dataset nehme2023textured.
• Figure 3: Crack maps of local windows of four different input pairs of reference and distorted frames obtained from the Nehmé et al.nehme2023textured (first two rows) and TSMD tsmd (last two rows) datasets. GMSD gmsd, SSIM wang2004image, and IW-SSIM iwssim quality maps are also generated and binarized with optimized thresholds and added as baselines for comparison with the proposed PCD method.
• Figure 4: Ablation study for contrast and Laplacian modulation components of the proposed PCD method for two sets of input pairs of reference and distorted frames obtained from the TSMD dataset tsmd.