Quadtree sequences improve pixel-level change detection in satellite images
From Pixels to Hierarchical Sequences: Quadtree Mask Encoding for Vision-Language Binary Change Detection
Computer Vision and Pattern Recognition
Summary
Detecting precise changes between two images taken at different times is important but challenging, especially for small or broken-up areas. The authors developed QUAKE-CD, a method that represents changes as compact tree-structured sequences that can be reliably checked and decoded. Their approach works better than existing methods by combining detailed visual clues with step-by-step reasoning. This helps create more accurate maps showing where changes have occurred in remote sensing images.
What this means in practice
- •For remote sensing engineers: Generate accurate pixel-level change maps from satellite images for environmental monitoring and urban planning.
- •For geospatial analysts: Use structured quadtree mask generation to improve detection of small or fragmented changes in paired aerial images.
Authors
Xiao An, Ruikang Zhang, Chen Zhong, Xuli Shen, Jiaxing Sun, Jiang Wu, Wei He
Abstract
Dense change detection in remote sensing requires vision-language models (VLMs) to compare bi-temporal images and generate accurate pixel-level masks. Existing VLMs are largely confined to change captioning outputs, and the few that produce pixel-level masks still rely on external decoders or flat text-as-mask serialization, which are less effective for small and fragmented changes. We introduce QUAKE-CD, a framework that recasts dense change prediction as syntax-verifiable structured generation. QUAKE-CD represents binary change masks as grammar-constrained quadtree token sequences, making the masks compact, syntactically checkable, and deterministically decodable within an autoregressive generation space. We further construct QUAKE-CoT, which pairs these sequences with chain-of-thought traces grounded in visual evidence, and jointly optimizes textual reasoning and spatial dense prediction through a progressive curriculum followed by grammar-gated dual-reward RL. On QUAKE-CoT, QUAKE-CD achieves 78.31% accumulated F1, outperforming decoder-based and flat text-as-mask VLMs while producing more faithful bi-temporal reasoning.