Point upsampling networks for single-photon sensing

2025-08-18

AI Summary

Single-photon sensing imaging offers long-range capability and ultra-high sensitivity, yet its output point clouds are inherently sparse and suffer from significant spatial distortion, severely limiting practical deployment. To address this, we propose the first deep learning-based upsampling framework specifically designed for single-photon point clouds. Our method innovatively integrates a multi-path scanning mechanism, a bidirectional Mamba backbone network, and an adaptive upsampling offset correction module within a unified architecture—jointly modeling global geometric structure and local fine-grained features. This design effectively enhances point density and corrects spatial distortions, enabling high-fidelity reconstruction. Extensive evaluations on standard benchmarks and real-world scenes demonstrate that our approach substantially outperforms state-of-the-art methods: it achieves higher reconstruction accuracy, superior noise robustness, richer geometric detail, improved visual consistency, and markedly reduced spatial distortion—establishing a novel paradigm toward practical single-photon imaging systems.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Single-photon sensing has generated great interest as a prominent technique of long-distance and ultra-sensitive imaging, however, it tends to yield sparse and spatially biased point clouds, thus limiting its practical utility. In this work, we propose using point upsampling networks to increase point density and reduce spatial distortion in single-photon point cloud. Particularly, our network is built on the state space model which integrates a multi-path scanning mechanism to enrich spatial context, a bidirectional Mamba backbone to capture global geometry and local details, and an adaptive upsample shift module to correct offset-induced distortions. Extensive experiments are implemented on commonly-used datasets to confirm its high reconstruction accuracy and strong robustness to the distortion noise, and also on real-world data to demonstrate that our model is able to generate visually consistent, detail-preserving, and noise suppressed point clouds. Our work is the first to establish the upsampling framework for single-photon sensing, and hence opens a new avenue for single-photon sensing and its practical applications in the downstreaming tasks.

Problem

• Upsampling sparse single-photon point clouds
• Reducing spatial distortion in point clouds
• Enhancing detail and noise suppression

Innovation

• Multi-path scanning enriches spatial context
• Bidirectional Mamba captures geometry details
• Adaptive upsample shift corrects distortions