Multi-level Edge Information Guided Progressive Network for Fine-Grained Cultivated Land Delineation

Fine-grained delineation of cultivated land parcels from optical remote sensing imagery is essential for agricultural monitoring and land resource management. However, existing approaches are often constrained by insufficient intra-parcel structural consistency and weak inter-class separability, resulting in fragmented predictions and blurred boundaries. To address these challenges, this work aims to improve the robustness of cultivated land delineation, with additional evaluation under varying spatial resolutions. A Multi-level Edge infoRmation Guided progrEssive Network (MergeNet) is developed to enhance both semantic completeness and boundary localization. MergeNet adopts Res2Net-50 as the backbone to extract hierarchical features and integrates a progressive edge-guided decoding strategy consisting of three functional modules. Specifically, an edge enhancement module (EEM) is designed to strengthen boundary sensitivity and provide complementary edge cues. Meanwhile, a structure-aware semantic enhancement module (SSEM) promotes semantic completeness through adaptive multi-scale feature aggregation. Furthermore, an edge-guided semantic refinement module (ESRM) progressively refines semantic features by integrating edge information derived from the input image, EEM, and SSEM, thereby improving boundary delineation. Deep supervision is applied at each decoding stage to facilitate network optimization and reinforce hierarchical feature learning. Exten-sive experiments conducted on three datasets with different spatial resolutions, including FGFD (0.3 m), iFLYTEK (0.75-1.1 m), and the Netherlands dataset (10 m), demonstrate that MergeNet achieves competitive performance in both pixel-level accuracy and structural quality. The consistent performance across different spatial conditions verifies the robustness of MergeNet for cultivated land delineation. The source code will be publicly available at https://github.com/lixinghua5540/MergeNet.