Research on Spectral Image Reconstruction Based on Nonlinear Spectral Dictionary Learning From Single RGB Image
ZUO Chu1, XIE De-hong2*, WAN Xiao-xia3
1. School of Light Industry and Food, Nanjing Forestry University, Nanjing 210037, China
2. College of Information Science and Technology, Nanjing Forestry University, Nanjing 210037, China
3. Hubei Province Engineering Technical Center for Digitization and Virtual Roprodcuction of Color Information of Culture Relics, Wuhan University, Wuhan 430079, China
Abstract:A nonlinear reconstruction method based on nonlinear spectral dictionary learning was proposed to solve the ill-posed problem of spectral image reconstruction from a single RGB image. In order to adapt to the linear and nonlinear data, the method firstly improves the nonlinear principal component analysis algorithm based on a modified self-association neural network model. It uses to learn the low-dimensional spectral dictionary from the training spectrum set, which is used in the inverse equation of spectral reconstruction to alleviate the ill condition. In addition, based on the spectral dictionary, the damped Gaussian Newton method combined with the truncated singular value decomposition regularization method is used further to alleviate the ill-posed problem of the nonlinear inversion, and the spectral image can be reconstructed from a single RGB image. In the experiment, two different spectral training sets, Munsell and Munsell+Pantone, were used to learn the spectral dictionary. Meanwhile, CAVE and UEA spectral image libraries were used for the spectral reconstruction tests. Compared with the existing methods, it is found that the average root means square error of CAVE and UEA spectral images reconstructed by this method under different spectral training sets were the lowest, which were 0.212 4, 0.255 4, 0.229 4 and 0.294 9 respectively. The standard deviations of root mean square error was close to the effect of the best method, which was 0.068 5, 0.084 7, 0.066 8 and 0.087 0 respectively. The results show that the method for reconstructing the spectral image from a single RGB image has advantages in accuracy and stability.
左 楚,谢德红,万晓霞. 基于非线性光谱字典学习的单幅RGB图像光谱重建方法研究[J]. 光谱学与光谱分析, 2022, 42(07): 2092-2100.
ZUO Chu, XIE De-hong, WAN Xiao-xia. Research on Spectral Image Reconstruction Based on Nonlinear Spectral Dictionary Learning From Single RGB Image. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2022, 42(07): 2092-2100.
[1] Khan M J, Khan H S, Yousaf A, et al. IEEE Access, 2018, 6: 14118.
[2] Kuhns T, Messinger D W. Spectral Phenomenology of Historical Parchments and Inks to Aid Cultural Heritage Imaging System Development. VelezReyes M, Messinger D W. Proceedings of SPIE,2018.
[3] Del Pozo S, Rodriguez-Gonzalvez P, Sanchez-Aparicio L J, et al. Multispectral Imaging in Cultural Heritage Conservation. Hayes J, Ouimet C, Quintero M S, et al. International Archives of the Photogrammetry, Remote Sensing and SpatialInformation Sciences. Gottingen: Copernicus Gesellschaft MBH, 2017. 155.
[4] He Y, Jiao W, Shi Y, et al. IEEE Journal of Biomedical and Health Informatics, 2020, 24(2): 493.
[5] Brossard M, Marion R, Carrere V. Remote Sensing Letters, 2016, 7(6): 581.
[6] Zazi L, Boutaleb A, Guettouche M S. Arabian Journal of Geosciences, 2017, 10(11): 252.
[7] Thomas J, Lapray P, Gouton P, et al. Sensors, 2016, 16(7): 993.
[8] Ji X, Miao C, Zhang Y, et al. Optics Communications, 2017, 404(SI): 11.
[9] Wu R, Li Y, Xie X, et al. Sensors, 2019, 19(13): 2905.
[10] Neumaier A. SIAM Rev., 1998, 40: 636.
[11] Chou T, Hsieh C, Chen E. Color Research and Application, 2019, 44(4): 508.
[12] Sibson R. A Brief Description of Natural Neighbor Interpolation, John Wiley & Sons, 1981.
[13] Cao B, Liao N, Li Y, et al. Optical Engineering, 2017, 56(5): 053104.
[14] Arad B, Ben-Shahar O. Sparse Recovery of Hyperspectral Signal from Natural RGB Images. Leibe B, Matas J, Sebe N, et al. Lecture Notes in Computer Science,2016, 9911: 19.
[15] Oh S W, Brown M S, Pollefeys M, et al. Do It Yourself Hyperspectral Imaging with Everyday Digital Cameras. IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE, 2016: 2461.
[16] Cao Q, Wan X, Li J, et al. Optical Review, 2016, 23(5): 753.
[17] Wu D, Tian J, Tang Y. Optical Review, 2014, 21(2): 117.
[18] Flinkman M, Laamanen H, Tuomela J, et al. Journal of the Optical Society of America A-Optics Image Science and Vision, 2013, 30(9): 1806.
[19] Chou Y, Cheung V, Luo M R, et al. Spectral Representation of Object Colours. Color and Imaging Conference Final Program and Proceedings,2012: 57.
[20] Rayat A, Amirshahi S H, Agahian F. Color Research and Application, 2014, 39(2): 136.
[21] Usui S, Nakauchi S, Nakano M. J. Opt. Soc. Am. A, 1992, 9(4): 516.
[22] LI Chan, WAN Xiao-xia, LIU Qiang, et al(李 婵,万晓霞,刘 强,等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2016, 36(5): 1400.
[23] Barakzehi M, Amirshahi S H, Peyvandi S, et al. Journal of the Optical Society of America A-Optics Image Science and Vision, 2013, 30(9): 1862.
[24] Licciardi G, Chanussot J. European Journal of Remote Sensing, 2018, 51(1): 375.
[25] Snee R D, Marquardt D W. The American Statistician, 1984, 38(2): 83.
[26] Hsieh W W. Tellus Series A-Dynamic Meteorology and Oceanography, 2001, 53(5): 599.
[27] Björck Å. Numerical Methods for Least Squares Problems, 1996. 339
[28] Hansen P C. BIT Numerical Mathematics, 1987, 27(4): 534.
[29] Deidda G P, Fenu C, Rodriguez G. Inverse Problems, 2014, 30(12): 125014.
[30] Armijo L. Pacific Journal of Mathematics, 1966, 16(1): 1.
[31] https://www.pantone.com/nylon-brights-set.
[32] Yasuma F, Mitsunaga T, Iso D, et al. IEEE Transactions on Image Processing, 2010, 19(9): 2241.
[33] Finlayson G D, Hordley S D, Morovic P. CGIV 2004: 2nd European Conference on Color in Graphics, Imaging, and Vision,2004. 268.