Reg-TTR, Test-Time Refinement for Fast, Robust and Accurate Image Registration

2026-01-27

AI Summary

Existing image registration methods struggle to simultaneously achieve speed, robustness, and accuracy: traditional approaches are robust but computationally slow, deep learning–based methods are efficient yet vulnerable to domain shift, and general-purpose foundation models often underperform task-specific ones in precision. This work proposes Reg-TTR, the first framework to incorporate test-time refinement into image registration. By seamlessly integrating a pretrained deep model with a classical iterative algorithm during inference, Reg-TTR efficiently refines initial alignment estimates. The method incurs only a 21% increase in inference time (+0.56 seconds) yet achieves state-of-the-art performance on two benchmark tasks, substantially narrowing the accuracy gap between general-purpose and specialized models while preserving near-deep-learning-level inference efficiency.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Multimodal Learning
• 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: Foundation models and LLMs for Web-related graphs

Abstract

Traditional image registration methods are robust but slow due to their iterative nature. While deep learning has accelerated inference, it often struggles with domain shifts. Emerging registration foundation models offer a balance of speed and robustness, yet typically cannot match the peak accuracy of specialized models trained on specific datasets. To mitigate this limitation, we propose Reg-TTR, a test-time refinement framework that synergizes the complementary strengths of both deep learning and conventional registration techniques. By refining the predictions of pre-trained models at inference, our method delivers significantly improved registration accuracy at a modest computational cost, requiring only 21% additional inference time (0.56s). We evaluate Reg-TTR on two distinct tasks and show that it achieves state-of-the-art (SOTA) performance while maintaining inference speeds close to previous deep learning methods. As foundation models continue to emerge, our framework offers an efficient strategy to narrow the performance gap between registration foundation models and SOTA methods trained on specialized datasets. The source code will be publicly available following the acceptance of this work.

Problem

• image registration
• accuracy
• foundation models
• domain shift
• inference speed

Innovation

• Test-Time Refinement
• Image Registration
• Foundation Models
• Deep Learning
• Domain Robustness