PanComplex: Leveraging Complex-Valued Neural Networks for Enhanced Pansharpening

Authors: Chunhui Luo, Dong Li, Xiaoliang Ma, Xin Lu, Zhiyuan Wang, Jiangtong Tan, Xueyang Fu

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on multiple datasets demonstrate that our method achieves optimal performance with the fewest parameters and exhibits strong generalization ability to other tasks.
Researcher Affiliation Collaboration Chunhui Luo1 , Dong Li1 , Xiaoliang Ma2 , Xin Lu1 , Zhiyuan Wang1 , Jiangtong Tan1 , Xueyang Fu1 1University of Science and Technology of China, Hefei, China 2Geovis, Hefei, China EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes the methodology through textual explanations and figures illustrating network architectures, but does not include explicit pseudocode or algorithm blocks.
Open Source Code Yes The source code for this work is publicly available at https://github.com/lch-ustc/Pan Complex.
Open Datasets Yes To evaluate the effectiveness of our network, we conduct experiments on three satellite datasets: World View-II (WV2), Gaofen2 (GF2) and World View-III (WV3). Each dataset contains a large number of paired low-resolution multispectral images and panchromatic images, which are divided into training, validation, and test sets. The dataset construction follows the methodology of previous studies, using the Wald protocol [Wald et al., 1997] tool to generate training and testing data. ... To further assess the generalization capability of our proposed method, we apply it to two additional tasks: depth image super-resolution using the NYU v2 dataset and infrared-RGB fusion on the Road Scene dataset.
Dataset Splits Yes Each dataset contains a large number of paired low-resolution multispectral images and panchromatic images, which are divided into training, validation, and test sets. ... Experiments are carried out on an additional 200 sets from the GF2 dataset.
Hardware Specification Yes In our experiments, all deep learning models are implemented using Py Torch and trained on an NVIDIA Ge Force GTX 3090 GPU.
Software Dependencies No The paper mentions "Py Torch" but does not specify a version number or other software with version numbers.
Experiment Setup Yes For each dataset, the multispectral (MS) images are cropped into patches of size 32 32, while the corresponding panchromatic (PAN) images are resized to 128 128. During the training phase, the networks are optimized using the Adam optimizer with an initial learning rate of 1 10 4 . After 200 epochs, the learning rate is reduced by half.