Unsupervised Band Selection Based on Group-Based Sparse Representation

  • Hung-Chang Chien
  • Chih-Hung Lai
  • Keng-Hao LiuEmail author
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10116)


Band selection (BS) is one of the important topics in hyperspectral image data analysis. How to search the representative bands that can effectively represent the image with lower inter-band redundancy is an long-term issue. Recently, the sparse representation (SR) was used to solve BS problem, called SR-BS. It aimed to find a set of representative bands that can represent the whole bands based on the minimization of reconstructed error in SR. However, those SR-BS methods suffer from an issue about higher complexity in the optimization process, even though the greedy strategy, such as orthogonal matching pursuit (OMP) algorithm, is used to accelerate them. Another issue is that the bands selected by SR-BS may not be really complementary since the homogeneity of adjacent spectral bands is not considered. To make SR-BS more efficient, in this paper, we model the BS problem as group sparse representation (GSR) problem where the dictionary matrix (i.e., all spectral bands) are pre-clustered to several non-overlapping groups based on the spectral similarity. Later, we adopt group orthogonal matching pursuit (GOMP) algorithm to solve the optimization problem. We named the proposed approach as GOMP-BS. Since GOMP-BS picks bands under group structure, it not only reaches higher computational efficiency but also makes the selected bands more less redundant. The experiments show that GOMP-BS achieves higher classification performance when the number of selected bands is low and requires less computation time than OMP-based methods.


Hyperspectral Image Land Cover Classification Image Scene Band Selection Overall Accuracy 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Chang, C.I.: Hyperspectral Imaging: Techniques for Spectral Detection and Classification, vol. 1. Springer, New York (2003)CrossRefGoogle Scholar
  2. 2.
    Chang, C.I.: Hyperspectral Data Processing: Algorithm Design and Analysis. Wiley, New York (2013)CrossRefzbMATHGoogle Scholar
  3. 3.
    Wright, J., Ma, Y., Mairal, J., Sapiro, G., Huang, T.S., Yan, S.: Sparse representation for computer vision and pattern recognition. Proc. IEEE 98, 1031–1044 (2010)CrossRefGoogle Scholar
  4. 4.
    Iordache, M.D., Bioucas-Dias, J.M., Plaza, A.: Sparse unmixing of hyperspectral data. IEEE Trans. Geosci. Remote Sens. 49, 2014–2039 (2011)CrossRefGoogle Scholar
  5. 5.
    Ehler, M., Hirn, M.: Sparse endmember extraction and demixing. In: IGARSS, pp. 1385–1388 (2012)Google Scholar
  6. 6.
    Li, S., Qi, H.: Sparse representation based band selection for hyperspectral images. In: 2011 18th IEEE International Conference on Image Processing, pp. 2693–2696. IEEE (2011)Google Scholar
  7. 7.
    Du, Q., Bioucas-Dias, J.M., Plaza, A.: Hyperspectral band selection using a collaborative sparse model. In: 2012 IEEE International Geoscience and Remote Sensing Symposium, pp. 3054–3057. IEEE (2012)Google Scholar
  8. 8.
    Elhamifar, E., Sapiro, G., Vidal, R.: See all by looking at a few: sparse modeling for finding representative objects. In: 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1600–1607. IEEE (2012)Google Scholar
  9. 9.
    Lai, C.H., Chen, C.S., Chen, S.Y., Liu, K.H.: Sequential band selection method based on group orthogonal matching pursuit. In: 8th Workshop on Hyperspectral Image and Signal Processing (Whispers 2016). IEEE (2016)Google Scholar
  10. 10.
    Sun, W., Zhang, L., Du, B.: A sparse self-representation method for band selection in hyperspectral. In: 7th Workshop on Hyperspectral Image and Signal Processing (Whispers 2015). IEEE (2015)Google Scholar
  11. 11.
    Chang, C.I., Liu, K.H.: Progressive band selection of spectral unmixing for hyperspectral imagery. IEEE Trans. Geosci. Remote Sens. 52, 2002–2017 (2014)CrossRefGoogle Scholar
  12. 12.
    Swirszcz, G., Abe, N., Lozano, A.C.: Grouped orthogonal matching pursuit for variable selection and prediction. In: Advances in Neural Information Processing Systems, pp. 1150–1158 (2009)Google Scholar
  13. 13.
    Donoho, D.L.: For most large underdetermined systems of linear equations the minimal \(\ell \)1-norm solution is also the sparsest solution. Commun. Pure Appl. Math. 59, 797–829 (2006)MathSciNetCrossRefzbMATHGoogle Scholar
  14. 14.
    Tropp, J.A., Gilbert, A.C.: Signal recovery from random measurements via orthogonal matching pursuit. IEEE Trans. Inf. Theory 53, 4655–4666 (2007)MathSciNetCrossRefzbMATHGoogle Scholar
  15. 15.
    Blumensath, T., Davies, M.E.: On the difference between orthogonal matching pursuit and orthogonal least squares (2007)Google Scholar
  16. 16.
    Yuan, M., Lin, Y.: Model selection and estimation in regression with grouped variables. J. Roy. Stat. Soc.: Ser. B (Stat. Methodol.) 68, 49–67 (2006)MathSciNetCrossRefzbMATHGoogle Scholar
  17. 17.
    Mojaradi, B., Emami, H., Varshosaz, M., Jamali, S.: A novel band selection method for hyperspectral data analysis. Int. Arch. Photogramm Remote Sens. Spat. Inf. Sci. 37, 447–454 (2008)Google Scholar
  18. 18.
    Su, H., Yang, H., Du, Q., Sheng, Y.: Semisupervised band clustering for dimensionality reduction of hyperspectral imagery. IEEE Geosci. Remote Sens. Lett. 8, 1135–1139 (2011)CrossRefGoogle Scholar
  19. 19.
    Chang, C.I., Wang, S.: Constrained band selection for hyperspectral imagery. IEEE Trans. Geosci. Remote Sens. 44, 1575–1585 (2006)CrossRefGoogle Scholar
  20. 20.
    Du, Q., Yang, H.: Similarity-based unsupervised band selection for hyperspectral image analysis. IEEE Geosci. Remote Sens. Lett. 5, 564–568 (2008)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing AG 2017

Authors and Affiliations

  1. 1.Department of Mechanical and Electro-Mechanical EngineeringNational Sun Yat-sen UniversityKaohsiungTaiwan

Personalised recommendations