Exploring Generalizable Pre-training for Real-world Change Detection via Geometric Estimation

2025-04-19

AI Summary

Manual registration of misaligned multi-temporal remote sensing images severely hampers automated change detection. Method: We propose MatchCD, an end-to-end self-supervised framework that jointly models geometric deformation estimation and semantic change detection. It introduces a geometry-estimation-driven self-supervised pretraining paradigm enabling direct end-to-end inference on native large-scale imagery (up to 6K×4K) without tiling or manual intervention. Contrastive learning acquires generalizable representations, which are then zero-shot transferred to the joint registration–change-detection optimization task. Contribution/Results: MatchCD achieves superior robustness and accuracy under severe geometric distortions and complex real-world conditions, fully eliminating the need for manual registration. It advances change detection toward unified, fully automated pipelines.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Unsupervised & Self-Supervised Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

As an essential procedure in earth observation system, change detection (CD) aims to reveal the spatial-temporal evolution of the observation regions. A key prerequisite for existing change detection algorithms is aligned geo-references between multi-temporal images by fine-grained registration. However, in the majority of real-world scenarios, a prior manual registration is required between the original images, which significantly increases the complexity of the CD workflow. In this paper, we proposed a self-supervision motivated CD framework with geometric estimation, called"MatchCD". Specifically, the proposed MatchCD framework utilizes the zero-shot capability to optimize the encoder with self-supervised contrastive representation, which is reused in the downstream image registration and change detection to simultaneously handle the bi-temporal unalignment and object change issues. Moreover, unlike the conventional change detection requiring segmenting the full-frame image into small patches, our MatchCD framework can directly process the original large-scale image (e.g., 6K*4K resolutions) with promising performance. The performance in multiple complex scenarios with significant geometric distortion demonstrates the effectiveness of our proposed framework.

Problem

• Addressing unaligned multi-temporal images in change detection
• Eliminating manual registration in real-world CD workflows
• Enabling direct large-scale image processing for change detection

Innovation

• Self-supervised contrastive representation for encoder optimization
• Zero-shot capability for image registration and change detection
• Direct processing of large-scale images without patching