Deep Learning-Based Point Cloud Registration: A Comprehensive Survey and Taxonomy

2024-04-22

AI Summary

Point cloud registration (PCR) lacks standardized evaluation protocols, hindering fair cross-method comparison of deep learning (DL) approaches under realistic conditions—namely, noise, outliers, and uncertain initial poses. To address this, we propose the first fine-grained taxonomy for DL-based PCR, systematically categorizing methods along four orthogonal dimensions: supervision paradigm (supervised vs. unsupervised), registration pipeline (end-to-end vs. feature-based), optimization strategy (differentiable vs. iterative), and network architecture (e.g., point-wise, graph-based, or transformer-based). We conduct the first standardized quantitative benchmark across 12 representative methods under unified data splits, training configurations, and evaluation metrics. We publicly release our comprehensive evaluation framework and benchmark results. Empirical analysis reveals fundamental performance boundaries and application-specific trade-offs, identifies critical bottlenecks—including sensitivity to large-angle initial misalignments and limited noise robustness—and provides principled guidance for algorithm design and rigorous evaluation.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Security and Privacy: Large-scale security measurements

Abstract

Point cloud registration involves determining a rigid transformation to align a source point cloud with a target point cloud. This alignment is fundamental in applications such as autonomous driving, robotics, and medical imaging, where precise spatial correspondence is essential. Deep learning has greatly advanced point cloud registration by providing robust and efficient methods that address the limitations of traditional approaches, including sensitivity to noise, outliers, and initialization. However, a well-constructed taxonomy for these methods is still lacking, making it difficult to systematically classify and compare the various approaches. In this paper, we present a comprehensive survey and taxonomy on deep learning-based point cloud registration (DL-PCR). We begin with a formal description of the point cloud registration problem, followed by an overview of the datasets, evaluation metrics, and loss functions commonly used in DL-PCR. Next, we categorize existing DL-PCR methods into supervised and unsupervised approaches, as they focus on significantly different key aspects. For supervised DL-PCR methods, we organize the discussion based on key aspects, including the registration procedure, optimization strategy, learning paradigm, network enhancement, and integration with traditional methods; For unsupervised DL-PCR methods, we classify them into correspondence-based and correspondence-free approaches, depending on whether they require explicit identification of point-to-point correspondences. To facilitate a more comprehensive and fair comparison, we conduct quantitative evaluations of all recent state-of-the-art approaches, using a unified training setting and consistent data partitioning strategy. Lastly, we highlight the open challenges and discuss potential directions for future study. A comprehensive collection is available at https://github.com/yxzhang15/PCR.

Problem

• Point Cloud Alignment
• Deep Learning
• Evaluation Standards

Innovation

• Deep Learning
• Point Cloud Registration
• Supervised and Unsupervised Learning