FACT: Multinomial Misalignment Classification for Point Cloud Registration

2025-04-09

AI Summary

This work addresses the challenging problem of evaluating registration quality for LiDAR point clouds. We propose the first framework that formulates registration error prediction as a multi-class misalignment classification task, enabling reliability assessment for diverse registrators—including ICP and GeoTransformer. Methodologically, we employ Point Transformer to extract local geometric features, design a regression-aware classification loss combining cross-entropy and Wasserstein distance, and augment training data with synthetically perturbed point cloud pairs. Evaluated on the CorAl benchmark, our approach significantly outperforms existing methods, achieving fine-grained discrimination across multiple misalignment severity levels—surpassing conventional residual-based and point cloud quality metrics. The framework is publicly available as open-source code and supports expert-in-the-loop refinement of registration outputs.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We present FACT, a method for predicting alignment quality (i.e., registration error) of registered lidar point cloud pairs. This is useful e.g. for quality assurance of large, automatically registered 3D models. FACT extracts local features from a registered pair and processes them with a point transformer-based network to predict a misalignment class. We generalize prior work that study binary alignment classification of registration errors, by recasting it as multinomial misalignment classification. To achieve this, we introduce a custom regression-by-classification loss function that combines the cross-entropy and Wasserstein losses, and demonstrate that it outperforms both direct regression and prior binary classification. FACT successfully classifies point-cloud pairs registered with both the classical ICP and GeoTransformer, while other choices, such as standard point-cloud-quality metrics and registration residuals are shown to be poor choices for predicting misalignment. On a synthetically perturbed point-cloud task introduced by the CorAl method, we show that FACT achieves substantially better performance than CorAl. Finally, we demonstrate how FACT can assist experts in correcting misaligned point-cloud maps. Our code is available at https://github.com/LudvigDillen/FACT_for_PCMC.

Problem

• Predicts alignment quality of registered lidar point clouds
• Generalizes binary alignment to multinomial misalignment classification
• Assists experts in correcting misaligned point-cloud maps

Innovation

• Uses point transformer-based network for feature processing
• Introduces custom regression-by-classification loss function
• Generalizes binary alignment to multinomial misalignment classification