隱密攻擊：透過密度導引幻象實現的魯棒性3D高斯潑濺毒化

StealthAttack: Robust 3D Gaussian Splatting Poisoning via Density-Guided Illusions

October 2, 2025

作者: Bo-Hsu Ke, You-Zhe Xie, Yu-Lun Liu, Wei-Chen Chiu

cs.AI

摘要

如神經輻射場（NeRF）和3D高斯濺射（3DGS）等3D場景表示方法，已顯著推進了新視角合成技術的發展。隨著這些方法的普及，解決其潛在脆弱性變得至關重要。我們分析了3DGS對圖像級別投毒攻擊的魯棒性，並提出了一種新穎的密度引導投毒方法。該方法通過核密度估計（KDE）識別低密度區域，策略性地注入高斯點，嵌入依賴視角的虛幻物體，這些物體在受污染的視角中清晰可見，而對無辜視角的影響則降至最低。此外，我們引入了一種自適應噪聲策略，以破壞多視圖一致性，進一步提升攻擊效果。我們提出了一種基於KDE的評估協議，系統性地評估攻擊難度，為未來研究提供客觀的基準測試。大量實驗證明，與現有頂尖技術相比，我們的方法展現出卓越的性能。項目頁面：https://hentci.github.io/stealthattack/

English

3D scene representation methods like Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) have significantly advanced novel view synthesis. As these methods become prevalent, addressing their vulnerabilities becomes critical. We analyze 3DGS robustness against image-level poisoning attacks and propose a novel density-guided poisoning method. Our method strategically injects Gaussian points into low-density regions identified via Kernel Density Estimation (KDE), embedding viewpoint-dependent illusory objects clearly visible from poisoned views while minimally affecting innocent views. Additionally, we introduce an adaptive noise strategy to disrupt multi-view consistency, further enhancing attack effectiveness. We propose a KDE-based evaluation protocol to assess attack difficulty systematically, enabling objective benchmarking for future research. Extensive experiments demonstrate our method's superior performance compared to state-of-the-art techniques. Project page: https://hentci.github.io/stealthattack/