Image Classification Based on Improved Spatial Pyramid Matching Model

  • Li FengEmail author
  • Xiaofeng Wang
  • Dongfang Chen
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10955)


Spatial pyramid matching (SPM) uses the statistics of local features in an image sub region as a global feature. It shows good performance in terms of generic image recognition. However, the disadvantages of this method are that the constructed visual dictionary is easy to fall into a local optimal solution due to the randomness of the initial centroid of k-means and it ignores the spatial distribution of salient object in images. In this research, we propose a new clustering method that using black hole algorithm to determine the initial center of k-means when constructing a visual dictionary and making the result have globally optimal solution and less computational costs. To better distinguish the target and background in the image, we propose discriminative SPM, which is a new representation that forms the image feature as a weighted sum of features over all pyramid levels. The weights are selected by the spatial distribution of salient objects in images. The resulting feature is compact and preserves high discriminative power. Thus reducing the effect of image background on classification. As documented in the experimental results, the proposed schemes can improve the classification accuracy of image compared to the other existing methods.


K-means Black hole algorithm Spatial pyramid Initial center Salient object Discriminative SPM 


  1. 1.
    Zheng, Z., Zhang, Y., Yan, L.: Global and local exploitation for saliency using bag-of-words. IET Comput. Vis. 8(4), 299–304 (2014)CrossRefGoogle Scholar
  2. 2.
    Wu, L., Hoi, S.C., Yu, N.: Semantics-preserving bag-of-words models and applications. IEEE Trans. Image Process. Publ. IEEE Sig. Process. Soc. 19(7), 1908–1920 (2010)MathSciNetCrossRefGoogle Scholar
  3. 3.
    Lazebnik, S., Schmid, C., Ponce, J.: Beyond bags of features: Spatial pyramid matching for recognizing natural scene categories. In: CVPR (2006)Google Scholar
  4. 4.
    Bosch, A., Zisserman, A., Munoz, X.: Representing shape with a spatial pyramid kernel. In: ACM International Conference on Image and Video Retrieval, pp. 401–408. ACM (2007)Google Scholar
  5. 5.
    Bosch, A., Zisserman, A., Munoz, X.: Image classification using random forests and ferns. In: ICCV (2007)Google Scholar
  6. 6.
    Yu, Q.: Optimization of initial clustering center selection using K-means algorithm. Appl. Comput. 5, 170–174 (2017)Google Scholar
  7. 7.
    Yang, J., Yu, K., Gong, Y., Huang, T.: Linear spatial pyramid matching using sparse coding for image classification, pp. 1794–1801 (2009)Google Scholar
  8. 8.
    Fu, H., Cao, X., Tu, Z.: Cluster-based co-saliency detection. IEEE Trans. Image Process. Publ. IEEE Sig. Process. Soc. 22(10), 3766 (2013)MathSciNetCrossRefGoogle Scholar
  9. 9.
    Underwood, G., Templeman, E., Lamming, L., Foulsham, T.: Is attention necessary for object identification? Evidence from eye movements during the inspection of real-world scenes. Conscious. Cogn. 17(1), 159–170 (2008)CrossRefGoogle Scholar
  10. 10.
    Hatamlou, A.: Black hole: a new heuristic optimization approach for data clustering. Inf. Sci. 222(3), 175–184 (2013)MathSciNetCrossRefGoogle Scholar
  11. 11.
    Huo, S., Zhou, Y., Lei, J., Ling, N., Hou, C.: Linear feedback control system based salient object detection. IEEE Trans. Multimed. 1(1), 1 (2017)CrossRefGoogle Scholar
  12. 12.
    Goferman, S., Zelnikmanor, L., Tal, A.: Context-aware saliency detection. IEEE Trans. Pattern Anal. Mach. Intell. 34(10), 1915–1926 (2012)CrossRefGoogle Scholar
  13. 13.
    Alamdar, F., Keyvanpour, M.R.: Effective browsing of image search results via diversified visual summarization by clustering and refining clusters. Sig. Image Video Process. 8(4), 699–721 (2014)CrossRefGoogle Scholar
  14. 14.
    Xing, H., Huang, C.: Feature extraction of raindrops based on ostu algorithm. Meteorological Science & Technology (2017)Google Scholar
  15. 15.
    Bouchekara, H.R.E.H.: Optimal power flow using black-hole-based optimization approach. Appl. Soft Comput. 24, 879–888 (2014)CrossRefGoogle Scholar
  16. 16.
    Yaghoobi, S., Hemayat, S.,Mojallali, H.: Image gray-level enhancement using Black Hole algorithm. In: International Conference on Pattern Recognition and Image Analysis, pp. 1–5. IEEE (2015)Google Scholar
  17. 17.
    Tong, W., Gao, X.W., Jiang, Z.J.: Parameters optimizing of lssvm based on black hole algorithm. J. Northeast. Univ. 35(2), 170–174 (2014)Google Scholar
  18. 18.
    Pashaei, E., Aydin, N.: Binary black hole algorithm for feature selection and classification on biological data. Appl. Soft Comput. 56, 94–106 (2017)CrossRefGoogle Scholar
  19. 19.
    Caicedo, Juan C., Cruz, Angel, Gonzalez, Fabio A.: Histopathology image classification using bag of features and kernel functions. In: Combi, Carlo, Shahar, Yuval, Abu-Hanna, Ameen (eds.) AIME 2009. LNCS (LNAI), vol. 5651, pp. 126–135. Springer, Heidelberg (2009). Scholar
  20. 20.
    Deselaers, T., Pimenidis, L., Ney, H.: Bag-of-visual-words models for adult image classification and filtering. In: International Conference on Pattern Recognition, pp. 1–4. IEEE (2008)Google Scholar
  21. 21.
    Li, F.F., Fergus, R., Perona, P.: Learning generative visual models from few training examples: an incremental bayesian approach tested on 101 object categories. Comput. Vis. Image Underst. 106(1), 178 (2007). Scholar
  22. 22.
    Griffin, G., Holub, A., Perona, P.: Caltech-256 object category dataset. California Institute of Technology (2007).

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Computer Science and TechnologyWuhan University of Science and TechnologyWuhanChina
  2. 2.Hubei Province Key Laboratory of Intelligent Information Processing and Real-Time IndustrialWuhanChina

Personalised recommendations