Graph Attention-Driven Bayesian Deep Unrolling for Dual-Peak Single-Photon Lidar Imaging

2025-04-03

AI Summary

This work addresses the challenge of 3D imaging with single-photon lidar under strong noise and multi-target-per-pixel conditions (e.g., bi-modal depth distributions). We propose the first interpretable dual-peak depth reconstruction method. Unlike conventional statistical approaches—limited in expressive power—or existing deep learning methods—lacking interpretability or assuming only unimodal depth—we introduce a hierarchical Bayesian modeling framework coupled with a graph attention–driven unrolled network architecture. Our design explicitly embeds physical priors into the network structure to enable dual-depth representation and per-pixel uncertainty quantification. By integrating Bayesian inference, geometric deep learning, and graph attention mechanisms, the method models structured point cloud features. Evaluated on both synthetic and real-world datasets, it achieves state-of-the-art accuracy, producing physically plausible depth estimates alongside calibrated uncertainty maps—marking the first interpretable reconstruction of single-pixel bi-modal returns.

Technology Category

• Computer Vision: Multi-modal Vision
• Reasoning under Uncertainty: Graphical Models
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Single-photon Lidar imaging offers a significant advantage in 3D imaging due to its high resolution and long-range capabilities, however it is challenging to apply in noisy environments with multiple targets per pixel. To tackle these challenges, several methods have been proposed. Statistical methods demonstrate interpretability on the inferred parameters, but they are often limited in their ability to handle complex scenes. Deep learning-based methods have shown superior performance in terms of accuracy and robustness, but they lack interpretability or they are limited to a single-peak per pixel. In this paper, we propose a deep unrolling algorithm for dual-peak single-photon Lidar imaging. We introduce a hierarchical Bayesian model for multiple targets and propose a neural network that unrolls the underlying statistical method. To support multiple targets, we adopt a dual depth maps representation and exploit geometric deep learning to extract features from the point cloud. The proposed method takes advantages of statistical methods and learning-based methods in terms of accuracy and quantifying uncertainty. The experimental results on synthetic and real data demonstrate the competitive performance when compared to existing methods, while also providing uncertainty information.

Problem

• Handling noisy environments in single-photon Lidar imaging
• Improving interpretability in deep learning-based Lidar methods
• Enabling dual-peak detection per pixel in Lidar imaging

Innovation

• Hierarchical Bayesian model for multiple targets
• Deep unrolling algorithm with neural network
• Dual depth maps with geometric deep learning