Author Identifier (ORCID)
Jianxin Li: https://orcid.org/0000-0002-9059-330X
Abstract
Hyperspectral target detection (HTD) aims at extracting targets from complex backgrounds while overcoming noise interference. Existing deep learning models for HTD usually suffer from low spatial resolution and unitary representation, especially in space-borne platforms. Super-resolution, as a critical technology to enhance the spatial details, could effectively address the aforementioned issue. To make super-resolution absolutely pose positive effects on target detection, this paper proposes an end-to-end novel super-resolution learning inspired spectral-spatial correlation network for hyperspectral target detection (SR-HTD) from the perspective of spatial and spectral regularization to achieve high-precision detection. Specifically, we designed a Spatial Correlation Aggregation (SCA) module inspired by latent low-rank representation to compensate for the absence of global spatial structure information inherent in the real scenario. This enables SR-HTD to reason about the underlying target scene and perform background estimation. Considering further enhancement and regularization of spectral and spatial information, we integrated the SR branches into the generative reconstruction module, focusing on the significance and structural differences between the two categories. In contrast to conventional models, the cascaded and complementary mechanisms fully benefit from HTD and SR, forming a more comprehensive solution and releasing barriers to target detection from a complex and noisy background. We generated synthetic panchromatic (PAN) images and hyperspectral images (HSIs) to evaluate the target detection performance on high-resolution images and the impact of SR on target detection. For most of the data sets tested, the detection accuracy of AUC(D,F) is the best and second-best among all the methods compared. Compared with state-of-the-art methods, the proposed method shows the expected superiority of the ABU and HYDICE datasets.
Keywords
Graph and generative learning, hyperspectral image processing, hyperspectral pansharpening, super-resolution, Target detection
Document Type
Journal Article
Date of Publication
1-1-2026
Publication Title
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Publisher
IEEE
School
School of Business and Law
Funders
Natural Science Basic Research Program of Shaanxi (2024JC-YBQN-0619) / Postdoctoral Research Project of Shaanxi Province (2023BSHYDZZ118, 2024BSHSDZZ089) / China Postdoctoral Science Foundation (2023M740366)
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Comments
Zhong, J., Li, Y., Li, J., Shi, Y., Xie, W., & Gamba, P. (2026). Super-resolution learning inspired spectral-spatial correlation network for hyperspectral target detection. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. Advance online publication. https://doi.org/10.1109/JSTARS.2026.3670937