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

TL;DR

This survey provides a comprehensive taxonomy and fair, unified evaluation of deep learning-based point cloud registration (DL-PCR), separating supervised and unsupervised approaches and detailing downstream procedures from descriptor extraction to pose estimation. It introduces a fine-grained organization of supervised methods by registration procedure, optimization strategy, learning paradigm, network design, and integration with traditional algorithms, while classifying unsupervised methods into correspondence-free and correspondence-based categories. The paper offers quantitative comparisons on common benchmarks (e.g., 3DMatch/3DLoMatch and KITTI) under unified settings, highlighting strengths and trade-offs across methods such as diffusion-based and transformer-based DL-PCR, as well as multimodal and pretraining strategies. Finally, it discusses open challenges—realistic data generation, multimodal fusion beyond images, vision-language integration, and semi-/unsupervised learning—and points to an open GitHub repository for further exploration and replication.

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.

Figures (8)

• Figure 1: A taxonomy of supervised DL-PCR algorithms. Methods published in the same year are grouped in curly brackets, with the superscripts $^{17}$, $^{18}$, $^{19}$, $^{20}$, $^{21}$, $^{22}$, $^{23}$, $^{24}$, and $^{25}$ indicate the publication years 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024, and 2025 respectively. Within each curly bracket, the listed methods are not in any particular chronological order.
• Figure 2: An overview of key registration procedures in supervised DL-PCR. Note that certain procedures, such as overlap prediction, similarity matrix optimization, and outlier filtering, may not be necessary for all registration methods.
• Figure 3: Illustration of DL-PCR methods using the following two optimization strategies: (a) GMM-based and (b) Multimodality-based.
• Figure 4: Illustration of DL-PCR methods using different learning paradigms: (a) contrastive learning, (b) meta learning, and (c) reinforcement learning.
• Figure 5: An example of Transformer-based DL-PCR.