基于多级循环优化网络的复杂光照高光谱反射率无参重建方法

doi:10.3964/j.issn.1000-0593(2025)08-2174-09

Abstract
References
Related Citation (15)

Download: PDF (31522 KB)
Export: BibTeX | EndNote (RIS)

Abstract Spectral reflectance reconstruction aims to calculate the true reflectance distribution of an object's surface in response to incident light from hyperspectral images. It is an important research topic in quantitative remote sensing and other hyperspectral imaging applications. Traditional reflectance reconstruction methods mainly rely on statistical models, often requiring stable flat-field illumination conditions or reference images as prior information for reflectance estimation. However, when reference images or other prior information are lacking in practical applications, especially under complex and varying lighting conditions, reconstruction accuracy using traditional methods is not ideal. This paper treats illumination estimation and reflectance reconstruction as a constrained matrix decomposition problem to address this issue. It proposes a multi-level cyclic optimization network model for spectral reflectance reconstruction. The model utilizes a hybrid channel-spatial attention mechanism to adaptively focus on key features in reflectance spectral images, thereby enhancing the extraction and amplification of critical information under non-uniform and multi-illuminant conditions, significantly improving reflectance reconstruction's robustness. Additionally, the network integrates a denoising mechanism comprising two modules: low-rank regularization and total variation regularization. The low-rank regularization module explores the intrinsic low-dimensional structures of illumination and reflectance to suppress noise interference. The total variation regularization module imposes spatial smoothness constraints on the reconstructed spectra, thereby improving reconstruction accuracy, reducing spectral mutations and redundant information, and ensuring spatial consistency throughout the process. To validate the effectiveness of the proposed method, this paper designs related data preprocessing, model training, and evaluation methods. The KAUST hyperspectral dataset is used as the training set in the training process, and different types of incident light source scenarios are simulated in the testing phase. Using the CIE 1964 10° standard observer color matching function as a reference, the hyperspectral images are converted into color images for visualization and quantitative performance evaluation. Experimental results show that the proposed reference-free reflectance reconstruction method outperforms traditional statistical-based reconstruction methods and current popular deep learning-based reconstruction methods regarding reconstruction accuracy indicators such as SAM and GFC. Particularly, without a calibrated whiteboard as a reference, the method still maintains high spectral reconstruction accuracy, demonstrating superior generalization capability and excellent reconstruction performance in complex lighting environments.

Key words：Spectral reflectance; Multi-level cyclic optimization; Hybrid attention mechanism; Reference-free reconstruction

Received: 2024-12-25 Accepted: 2025-04-08

ZTFLH:

O433.4

Corresponding Authors: GAO Kun E-mail: gaokun@bit.edu.cn

Cite this article:

ZHANG He,GAO Kun,KE Kun-xin, et al. Reference-Free Reconstruction of Hyperspectral Reflectance Under Complex Illumination Based on Multi-Level Cyclic Optimization Network[J]. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(08): 2174-2182.

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2025)08-2174-09 OR https://www.gpxygpfx.com/EN/Y2025/V45/I08/2174

[1] Dib A, Thebault C, Ahn J, et al. Towards High Fidelity Monocular Face Reconstruction with Rich Reflectance Using Self-Supervised Learning and Ray Tracing, Proceedings of the IEEE/CVF International Conference on Computer Vision，2021: 12819.
[2] Ren X, Deng J, Cheng Y, et al. Monocular Identity-Conditioned Facial Reflectance Reconstruction. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 885.
[3] Wang W Z, Deng N, Xin B J. Optik, 2020, 208: 164491.
[4] Gong L X, Zhu C X, Luo Y E, et al. Applied Spectroscopy, 2023, 77(2): 200.
[5] Kınlı F, Özcan B, Kiraç F. Journal of Visual Communication and Image Representation, 2025, 108: 104412.
[6] Celik T, Yetgin Z. Signal, Image and Video Processing, 2015, 9: 1889.
[7] Wang L X, Sole A, Hardeberg J Y, et al. Optics Express, 2021, 29(16): 24695.
[8] Ulucan O, Ulucan D, Ebner M. Investigating Color Illusions from the Perspective of Computional Color Constancy, 2023, arXiv: 2312. 13114.
[9] Sippel F, Seiler J, Genser N, et al. Journal of the Optical Society of America A-Optics，Image Science and Vision, 2020, 37(11): 1695.
[10] WANG Han, LI Chang-jun(王晗，李长军). Laser and Optoelectronics Progress(激光与光电子学进展), 2022, 59(19): 1933001.
[11] Vršnak D, Domislovic′ I, Subašic′ M, et al. IEEE Access, 2023, 11: 2128.
[12] Lazar M, Hladnik A. Sensors, 2023, 23(2): 1000.
[13] Robles-Kelly A, Wei R. A Convolutional Neural Network for Pixelwise Illuminant Recovery in Color and Spectral Images, International Conference on Pattern Recognition, 2018, 109.
[14] Xu B, Liu J, Hou X, et al. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2020: 3616.
[15] Bianco S, Cusano C, Schettini R. IEEE Transactions on Image Processing, 2017, 26(9): 4347.
[16] Xie M D, Zhong C Z, He Y L, et al. 2025, arXiv: 2502. 12418.
[17] Afifi M, Hu Z H, Liang L. Optimizing Illuminant Estimation in Dual-Exposure HDR Imaging, in Computer Vision—ECCV 2024, Vol 15060, Springer, 2024.
[18] Li C X, Kang X J, Zhang Z F, et al. SWBNet: A Stable White Balance Network for sRGB Images, Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 1278.
[19] Xu J H, Luo M R, Fan H. Color Research and Application, 2023, 48(4): 368.
[20] Woo S, Park J, Lee J Y. CBAM: Convolution Block Attention Module, Proceedings of the European Conference on Computer Vision，2018: 3.
[21] MA Xiang-cai, CAO Qian, BAI Chun-yan, et al(麻祥才, 曹前, 白春燕, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2024, 44(3): 610.
[22] Li Y Q, Fu Q, Heidrich W. Multispectral Illumination Estimation Using Deep Unrolling Network, IEEE Conference on Computer Vision, 2021, 12672.
[23] Bianco S, Cusano C, Schettini R. Color Constancy Using CNNs, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, 2015: 81.
[24] Hu Y M, Wang B Y, Lin S. FC4: Fully Convolutional Color Constance With Confidence-Weighted Pooling. IEEE Conference on Computer Vision and Pattern Recognition, 2017, 4085.
[25] Entok U C, Laakom F, Pakdaman F, et al. Pixel-Wise Color Constancy via Smoothness Techniques in Multi-Illuminant Scenes, IEEE International Conference on Image Processing, 2024: 2737.
[26] Du S S, Zhao D R, Guan L L, et al. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 4415514.
[27] Liang C, Zhao E L, Ma L. Proceedings of SPIE, 2024, 13227: 1322702.