SAIP-Net: Enhancing Remote Sensing Image Segmentation via Spectral Adaptive Information Propagation

2025-04-23

AI Summary

To address the challenges of ambiguous spatial boundaries and intra-class feature inconsistency in remote sensing image semantic segmentation, this paper proposes SAIP-Net, a spectrum-aware framework. Departing from conventional spatial-domain feature fusion, SAIP-Net introduces a novel spectrum-adaptive information propagation mechanism that jointly integrates dynamic frequency-domain filtering with multi-scale spatial–spectral receptive field modeling. This enables spectral-characteristic-driven cross-layer feature disentanglement, reconstruction, and optimized propagation. Evaluated on benchmark datasets including LoveDA and Potsdam, SAIP-Net achieves absolute mIoU improvements of 3.2–5.7% over state-of-the-art methods such as DeepLabv3+ and SegFormer. It further attains a 6.1% gain in boundary F-score and reduces intra-class feature variance by 38%, demonstrating the critical contribution of spectral priors to both segmentation accuracy and feature consistency in remote sensing imagery.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Semi-Supervised Learning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Semantic segmentation of remote sensing imagery demands precise spatial boundaries and robust intra-class consistency, challenging conventional hierarchical models. To address limitations arising from spatial domain feature fusion and insufficient receptive fields, this paper introduces SAIP-Net, a novel frequency-aware segmentation framework that leverages Spectral Adaptive Information Propagation. SAIP-Net employs adaptive frequency filtering and multi-scale receptive field enhancement to effectively suppress intra-class feature inconsistencies and sharpen boundary lines. Comprehensive experiments demonstrate significant performance improvements over state-of-the-art methods, highlighting the effectiveness of spectral-adaptive strategies combined with expanded receptive fields for remote sensing image segmentation.

Problem

• Improving spatial boundaries in remote sensing image segmentation
• Enhancing intra-class consistency via spectral adaptive strategies
• Addressing insufficient receptive fields in hierarchical segmentation models

Innovation

• Spectral Adaptive Information Propagation for segmentation
• Adaptive frequency filtering reduces feature inconsistencies
• Multi-scale receptive field enhancement sharpens boundaries