MPEFNet: Multi-level Progressive Enhancement Fusion Network for Pansharpening

Remote sensing image fusion is a key technique to fuse low spatial resolution multispectral (MS) images with high panchromatic (PAN) obtain (HRMS) images. However, many existing algorithms typically perform single upsampling on the MS match its that of PAN image, and subsequently output fused through steps feature extraction, fusion, decoding. This single-stage approach not only fails fully utilize low-frequency high-frequency information in but also leads inadequate extraction internal spec-tral original resulting problems such as blurring, artifacts, incomplete spectral recovery image. To address these issues, this paper proposed multi-level progressive enhancement network (MPEFNet). different images, employs three-stage structure. The High Preserving Block (HPB) used alleviate detail distortion loss caused by upsampling. Bands Aggregation Module (BAM) Spatial (SAM) are refine module's features. Mean-while, Enhanced Fusion (EFM) further performs self-enhancement refined features, well mutu-al-enhancement information. method superior comparison qualitative analysis quantitative IKONOS WorldView-2 (WV-2) datasets.