Integrated Dynamic Phenological Feature for Remote Sensing Image Land Cover Change Detection

TL;DR

Phenological variability in natural land covers creates pseudo-changes that challenge remote sensing change detection. The InPhea framework combines a detector with a constrainer to learn phenology-aware change representation; the detector uses a differential attention module, twin high-resolution feature extractors, and a spatial pyramid block, while the constrainer imposes four priors and uses multi-stage contrastive learning to align features across phenological stages. It introduces a novel differential attention module, SPB, and four constraint modules (CDM, SCM, CLEM, PLM), plus a PSCD-Wuhan dataset, with extensive ablations showing performance gains and phenology robustness. Across three datasets (HRSCD, SECD, PSCD-Wuhan), InPhea outperforms state-of-the-art methods in IoU and F1, demonstrating effective mitigation of phenology-induced pseudo-changes and improved change-detection reliability.

Abstract

Remote sensing image change detection (CD) is essential for analyzing land surface changes over time, with a significant challenge being the differentiation of actual changes from complex scenes while filtering out pseudo-changes. A primary contributor to this challenge is the intra-class dynamic changes due to phenological characteristics in natural areas. To overcome this, we introduce the InPhea model, which integrates phenological features into a remote sensing image CD framework. The model features a detector with a differential attention module for improved feature representation of change information, coupled with high-resolution feature extraction and spatial pyramid blocks to enhance performance. Additionally, a constrainer with four constraint modules and a multi-stage contrastive learning approach is employed to aid in the model's understanding of phenological characteristics. Experiments on the HRSCD, SECD, and PSCD-Wuhan datasets reveal that InPhea outperforms other models, confirming its effectiveness in addressing phenological pseudo-changes and its overall model superiority.

Figures (6)

• Figure 1: The overall architecture of InPhea.
• Figure 2: Differential Attention Module Diagram.
• Figure 3: Spatial Pyramid Block Structure Diagram.
• Figure 4: CLEM and PLM Structure Diagram.
• Figure 5: Constrainer-assisted Multi-stage Contrastive Learning Schematic.