CORTEXA
← Browse
arxivcs.CV2026-06-30

Symmetry-Structured Neural Completion of Islamic Geometric Patterns from Sparse Control Geometry

Hassan Ugail, Irfan Mehmood

Islamic geometric patterns are governed by exact rotational symmetry and strict construction rules. This paper treats these rules as formal geometric knowledge and embeds them in a neural completion framework, rather than leaving them to be learned statistically from data. Given sparse control geometry and a target symmetry order, the system completes the pattern as a vector graph by predicting edges and refinements of bounded curves over a candidate lattice whose edges are organised into rotational orbits under the cyclic group. Symmetry is enforced either by constraining predictions within these orbits or by projecting them onto them during inference. The orbit-tied variant provides a constructive guarantee: for any input and any orbit-level selection rule, it produces exact N-fold symmetry, preserves anchor points, and keeps all refinements within prescribed bounds. These properties are verified numerically. The study focuses on rotational symmetry, and all quantitative results are obtained from procedurally generated graphs inspired by Islamic geometric design rather than from a historical corpus. On clean inputs, enforcing exact validity produces no measurable loss in fidelity. When control geometry is missing, an unstructured decoder loses fidelity and breaks symmetry; retraining on corrupted inputs recovers much of the fidelity but not exact validity. Symmetry-structured inference, by contrast, keeps violations at zero throughout. The results show that augmentation and symmetry structure address distinct failure modes: augmentation improves fidelity under corruption, while symmetry structure guarantees validity. The framework therefore provides a knowledge-constrained, guarantee-backed approach to neural completion for scalable vector ornaments whose validity depends on exact geometric structure.

View free PDFSource page

Related papers

arxivcs.CVcs.AI2026-07-06

Targeted Structure Completion for Sparse-View 3D Reconstruction in Autonomous Driving

Guoqing Wang, Pin Tang, Xiangxuan Ren, Liping Hou, Chao Ma

Reconstructing 3D scene structures from sparse, low-overlap observations remains a fundamental challenge in autonomous driving. Recent state-of-the-art frameworks achieve promising results by incorporating voxel-based Gaussians, but incur substantial computational redundancy due…

View free PDFSource page
arxivcs.CV2026-07-24

fMRI2Face: A Full-HD fMRI-Video Dataset and Geometry-Guided Neural Decoding Framework for Dynamic Human Face Reconstruction

Jingyang Huo, Xiangru Huang, Chentao Shen, Yikai Wang, Yun Wang, Jianxiong Gao, et al.

Reconstructing dynamic human faces from brain activity provides a powerful way to study how the mind perceives identity, expression, and facial motion. However, progress in fMRI-based face decoding has been limited by scarce controlled, high-resolution neural datasets and by meth…

View free PDFSource page
arxivcs.CV2026-07-06

SparseOcc++: Geometry-Aware Sparse Latent Representation for Semantic Occupancy Prediction

Pin Tang, Zhongdao Wang, Guoqing Wang, Xiangxuan Ren, Chao Ma

Vision-based 3D semantic occupancy prediction is essential for autonomous driving, yet dense voxel representations waste computation on largely empty space, while BEV and TPV projections compromise fine-grained 3D structure. Fully sparse representations offer an attractive altern…

View free PDFSource page
arxivcs.LGcs.CV2026-07-03

Observable- and Positional-Encoding-Dependent Symmetry Readout from Neural Network Weights

Naoya Chiba, Satoshi Sugiyama, Yuki Uranishi

Post-hoc analysis of trained neural network weights often seeks to recover geometric structure directly from the parameters. We show that, for positional-encoding-equipped neural fields, the symmetry visible from weights is not the true symmetry group itself, but an observable sy…

View free PDFSource page
arxivcs.CVcs.GR2026-06-27

Reliability-Aware Monocular Depth Supervision for Sparse-View Neural Reconstruction

Wei-Teng Chu, Yashasvini Gopalan, Changju Yuan

Sparse-view neural reconstruction is challenging in outdoor driving scenes, where cameras usually move along a narrow forward-facing trajectory and provide limited multi-view overlap. Although monocular depth estimators can provide dense geometric priors, their predictions are no…

View free PDFSource page
arxivcs.CVcs.AIcs.RO2026-07-24

SM4RT: Learning Structured Motion Geometry for 4D Reconstruction

Shing Ho J. Lin, Wenzhao Zheng, Dong Zhuo, Yuqi Wu, Jie Zhou, Jiwen Lu

Geometry Foundation Models (GFMs) have substantially advanced monocular 3D reconstruction, yet extending this capability to 4D dynamic understanding remains a fundamental challenge. Most existing motion perception methods (e.g., sparse tracking, dense point-wise flow) treat motio…

View free PDFSource page