Depth-Guided Privacy-Preserving Visual Localization Using 3D Sphere Clouds

2026-05-01

AI Summary

This work addresses the privacy leakage risks inherent in high-fidelity reconstruction for visual localization using sparse 3D point clouds, particularly the vulnerability of existing line-cloud representations to density-based attacks. To mitigate these issues, we propose a novel privacy-preserving scene representation termed the sphere cloud, which elevates map points to 3D rays passing through the scene centroid to effectively mislead density attacks. A lightweight cloud construction strategy is introduced to defend against image recovery attacks, and absolute depth maps from ToF sensors are integrated to resolve translational scale ambiguity in pose estimation. Experiments on public RGB-D datasets demonstrate that the sphere cloud achieves a favorable balance among real-time performance, localization accuracy, and robust privacy protection, outperforming current depth-guided localization approaches.

Abstract

The emergence of deep neural networks capable of revealing high-fidelity scene details from sparse 3D point clouds has raised significant privacy concerns in visual localization involving private maps. Lifting map points to randomly oriented 3D lines is a well-known approach for obstructing undesired recovery of the scene images, but these lines are vulnerable to a density-based attack that can recover the point cloud geometry by observing the neighborhood statistics of lines. With the aim of nullifying this attack, we present a new privacy-preserving scene representation called \emph{sphere cloud}, which is constructed by lifting all points to 3D lines crossing the centroid of the map, resembling points on the unit sphere. Since lines are most dense at the map centroid, the sphere cloud mislead the density-based attack algorithm to incorrectly yield points at the centroid, effectively neutralizing the attack. Nevertheless, this advantage comes at the cost of i) a new type of attack that may directly recover images from this cloud representation and ii) unresolved translation scale for camera pose estimation. To address these issues, we introduce a simple yet effective cloud construction strategy to thwart new attack and propose an efficient localization framework to guide the translation scale by utilizing absolute depth maps acquired from on-device time-of-flight (ToF) sensors. Experimental results on public RGB-D datasets demonstrate sphere cloud achieves competitive privacy-preserving ability and localization runtime while not excessively compensating the pose estimation accuracy compared to other depth-guided localization methods.

Problem

• visual localization
• privacy preservation
• 3D point clouds
• density-based attack
• scene reconstruction

Innovation

• sphere cloud
• privacy-preserving localization
• density-based attack
• depth-guided pose estimation
• 3D line representation