Searcharxiv⌕ Search

arXiv subjects

Yishai Lavi

Publications and source records attributed to Yishai Lavi.

4 recordsLinked to original sources

Propose & Verify: Fast Reflection Symmetry Detection via Harmonic Fields

Detecting multiple reflection symmetries in 3D shapes requires aggregating geometric evidence over a large space of candidate planes, making traditional voting approaches computationally expensive. We introduce the Fourier Accelerated Symmetry Transform (FAST), a training-free framework that combines geometric pair voting with efficient spectral computation. Exploiting the fact that global reflection planes pass through the shape centroid, we reduce the search to plane orientations, while geometric consistency filtering and independent witnesses suppress unreliable pair hypotheses. Our key insight is to represent these votes as a signal over plane orientations and efficiently compute a continuous, smoothed symmetry field in the spherical-harmonic domain. Specifically, we exploit the Funk-Hecke theorem, the spherical counterpart of the Fourier convolution theorem for rotationally symmetric kernels, to transform Gaussian smoothing into multiplication of harmonic coefficients. This enables efficient field evaluation, from which symmetry candidates are extracted, refined, and geometrically verified. Across two benchmarks, including a new benchmark of highly symmetric shapes, FAST achieves the highest clean multi-plane detection $F_1$ scores among evaluated methods (0.861 and 0.947), with GPU inference below 4 ms per shape: over $100\times$ faster than the next-fastest evaluated multi-plane GPU baseline. FAST also demonstrates strong robustness to geometric noise and competitive single-plane detection, achieving 94.9% success on clean single-plane shapes.

cs.CV↗

The spectrum of free projective geometries over finite chain rings

Let $\mathcal{O}$ be a Dedekind domain, $\mathfrak{p}$ a prime ideal of residue order $q$, and $\mathcal{O}_r=\mathcal{O}/\mathfrak{p}^r$ which is a finite chain ring; the motivating example is $\mathbb{Z}/p^r$. The free projective geometry $\mathbb{P}^{d-1}_{fr}(\mathcal{O}_r)$ is the graph whose vertices are the nontrivial free submodules of $\mathcal{O}_r^{d}$, two of them adjacent when one strictly contains the other. For every $d$, $q$ and $r$ we reduce its adjacency matrix to explicit symmetric matrices (blocks) of size at most $d-1$. There is one block for each signature (a partition with at most $d/2$ parts, each at most $r$) and we show that every block, at every depth, is a single universal matrix $L(s,σ)$ evaluated at the number $s$ of parts of the signature $λ$ and at $σ=2rs-|λ|$. The same blocks give the spectrum of the subgraph induced on any set of levels, and for two levels this spectrum is completely explicit. All of it depends on the ring only through $q$ and $r$; in particular the subgraphs $\mathbb{P}^{d-1}_{S}$ over $\mathbb{Z}/p^r$ and over $\mathbb{F}_p[t]/(t^r)$ are isospectral for all $\varnothing\ne S\subseteq\{1,\dots,d-1\}$; together with known rigidity results of arXiv:2510.09324, this proves that they are isospectral and non-isomorphic whenever $S$ contains at least two levels, as conjectured there.

math.CO↗

Frequency-Aware Gaussian Splatting Decomposition

3D Gaussian Splatting (3D-GS) enables efficient novel view synthesis, but treats all frequencies uniformly, making it difficult to separate coarse structure from fine detail. Recent works have started to exploit frequency signals, but lack explicit frequency decomposition of the 3D representation itself. We propose a frequency-aware decomposition that organizes 3D Gaussians into groups corresponding to Laplacian-pyramid subbands of the input images. Each group is trained with spatial frequency regularization to confine it to its target frequency, while higher-frequency bands use signed residual colors to capture fine details that may be missed by lower-frequency reconstructions. A progressive coarse-to-fine training schedule stabilizes the decomposition. Our method achieves state-of-the-art reconstruction quality and rendering speed among all LOD-capable methods. In addition to improved interpretability, our method enables dynamic level-of-detail rendering, progressive streaming, foveated rendering, promptable 3D focus, and artistic filtering. Our code will be made publicly available.

cs.CV↗

Geometry over finite local rings: Rigidity and Isospectrality

We study the simplicial order complexes obtained from free modules over finite local rings. These complexes arise naturally as geodesic spheres in Bruhat-Tits buildings over non-archimedean local fields. We establish two forms of rigidity, showing that their automorphism groups arise from the underlying algebraic group, and that they are determined by sparse induced subgraphs. We compute the spectra of these subgraphs and show that they form excellent expanders, which results in expansion for geodesic powers of Bruhat-Tits buildings. The computation also reveals that local rings with the same residue order give rise to isospectral induced subgraphs. Combining this with our rigidity results we show that the graphs arising from $n$-spaces over $\mathbb{Z}/p^{r}$ and $\mathbb{F}_{p}[t]/(t^{r})$ are isospectral and non-isomorphic.

math.GR↗