Transfer Learning With Nonlinear Spectral Synthesis for Hyperspectral Target Detection
- 30 November 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 61 (01962892) , 1-17
Abstract
Spectral distortion severely limits detection performance in hyperspectral imagery, while feature learning with neural networks could provide sufficient capacity to enhance spectral consistency. This paper designs an end-to-end hyperspectral target detection (HTD) network based on transfer learning and nonlinear spectral synthesis (TLNSS). We first utilize bilinear mixture model (BMM) to synthesize nonlinear target and background spectra for training sample augmentation, which could better characterize ground objects in complex environments. Due to the mutual constraints between the quantity and diversity of the synthesized spectra, transfer learning is introduced to further address data insufficiency. Specifically, we propose an asymmetric autoencoder with a particularly designed multi-level loss to maximally distinguish the reconstruction residuals of background and target, where the multi-scale feature extraction sub-network is trained with abundant reference data, and the simple restoration sub-network is updated with the simulated spectra. To effectively reconstruct the input as expected, the features extracted from different blocks are complementarily integrated through residual attention. Lastly, we accumulate reconstruction residuals across all levels for final detection. The experimental results and ablation analysis of single-data detection on three hyperspectral images verify the superiority and effectiveness of the proposed method, and further cross-data detection consolidates the satisfactory tolerance of TLNSS to spectral variation.Keywords
Funding Information
- National Nature Science Foundation of China (12171234)
- Fundamental Research Funds for the Central Universities, Chang’an University (300102223104)
- Postdoctoral Research Project of Shaanxi Province (2023BSHYDZZ118)
- China Postdoctoral Science Foundation (2023M740366)
This publication has 50 references indexed in Scilit:
- Hyperspectral Airborne “Viareggio 2013 Trial” Data Collection for Detection Algorithm AssessmentIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016
- Combined sparse and collaborative representation for hyperspectral target detectionPattern Recognition, 2015
- Hierarchical Suppression Method for Hyperspectral Target DetectionIEEE Transactions on Geoscience and Remote Sensing, 2015
- A Review of Nonlinear Hyperspectral Unmixing MethodsIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014
- An Automatic Robust Iteratively Reweighted Unstructured Detector for Hyperspectral ImageryIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014
- Application of the Hyperspectral Imager for the Coastal Ocean to Phytoplankton Ecology Studies in Monterey Bay, CA, USARemote Sensing, 2014
- Hyperspectral Imager for the Coastal Ocean: instrument description and first imagesApplied Optics, 2011
- Exploiting Spectral and Spatial Information in Hyperspectral Urban Data With High ResolutionIEEE Geoscience and Remote Sensing Letters, 2004
- Automatic spectral target recognition in hyperspectral imageryIEEE Transactions on Aerospace and Electronic Systems, 2003
- The CFAR adaptive subspace detector is a scale-invariant GLRTIEEE Transactions on Signal Processing, 1999