Approximation of Linear Discriminant Analysis for Word Dependent Visual Features Selection

  • Hervé Glotin
  • Sabrina Tollari
  • Pascale Giraudet
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3708)


To automatically determine a set of keywords that describes the content of a given image is a difficult problem, because of (i) the huge dimension number of the visual space and (ii) the unsolved object segmentation problem. Therefore, in order to solve matter (i), we present a novel method based on an Approximation of Linear Discriminant Analysis (ALDA) from the theoretical and practical point of view. Application of ALDA is more generic than usual LDA because it doesn’t require explicit class labelling of each training sample, and however allows efficient estimation of the visual features discrimination power. This is particularly interesting because of (ii) and the expensive manually object segmentation and labelling tasks on large visual database. In first step of ALDA, for each word w k , the train set is split in two, according if images are labelled or not by w k . Then, under weak assumptions, we show theoretically that Between and Within variances of these two sets are giving good estimates of the best discriminative features for w k . Experimentations are conducted on COREL database, showing an efficient word adaptive feature selection, and a great enhancement (+37%) of an image Hierarchical Ascendant Classification (HAC) for which ALDA saves also computational cost reducing by 90% the visual features space.


feature selection Fisher LDA visual segmentation image auto-annotation high dimension problem word prediction CBIR HAC COREL database PCA 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Barnard, K., Duygulu, P., Freitas, N., Forsyth, D., Blei, D., Jordan, M.: Matching words and pictures. Jour. of Machine Learning Research 3 (2003)Google Scholar
  2. 2.
    Barnard, K., Duygulu, P., Guru, R., Gabbur, P., Forsyth, D.: The effects of segmentation and feature choice in a translation model of object recognition. In: Computer Vision and Pattern Recognition, pp. 675–682 (2003)Google Scholar
  3. 3.
    Duda, R., Hart, P., Stork, D.: Pattern Classification. Wiley, Chichester (2000)Google Scholar
  4. 4.
    Luettin, J., Potamianos, C.N.G.: Hierarchical discriminant features for audio-visual LVCSR. In: Proc. of IEEE Int. Conf. ASSP (2001)Google Scholar
  5. 5.
    Glotin, H., Tollari, S.: Fast image auto-annotation with visual vector approximation clusters. In: IEEE EURASIP Content-Based Multimedia Indexing (2005)Google Scholar
  6. 6.
    Gosselin, P., Cord, M.: A comparison of active classification methods for content-based image retrieval. In: Proc. CVDB04 with SIGMOD 2004, Paris (2004)Google Scholar
  7. 7.
    Liu, Q., Huang, R., Lu, H., Ma, S.: Face recognition using kernel based Fisher discriminant analysis. In: Proc. of Automatic Face & Gesture Recognition (2002)Google Scholar
  8. 8.
    Monay, F., Gatica-Perez, D.: On image auto-annotation with latent space models. In: Proc. ACM Int. Conf. on Multimedia (ACM MM), pp. 275–278 (2003)Google Scholar
  9. 9.
    Muller, H., Marchand-Maillet, S., Pun, T.: The truth about corel - evaluation in image retrieval. In: Lew, M., Sebe, N., Eakins, J.P. (eds.) CIVR 2002. LNCS, vol. 2383, Springer, Heidelberg (2002)CrossRefGoogle Scholar
  10. 10.
    Neti, C., Potamianos, G., Luettin, J., Matthews, I., Glotin, H., Vergyri, D.: Large-vocabulary audio-visual speech recognition: A summary of the J. Hopkins Summer 2000 Wksp. In: IEEE Wksp. Multimedia Signal Process. (2001)Google Scholar
  11. 11.
    Shi, J., Malik, J.: Normalized cuts and image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence 22(8), 888–905 (2000)CrossRefGoogle Scholar
  12. 12.
    Tollari, S., Glotin, H.: Keyword dependant selection of visual features and their heterogeneity for image content-based interpretation. Technical Report LSIS.RR.2005.003, LSIS, Similar content submitted to ACMMM 2005 (2005)Google Scholar
  13. 13.
    Tollari, S., Glotin, H., Le Maitre, J.: Enhancement of textual images classification using segmented visual contents for image search engine. Multimedia Tools and Applications 25(3), 405–417 (2005)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2005

Authors and Affiliations

  • Hervé Glotin
    • 1
  • Sabrina Tollari
    • 1
  • Pascale Giraudet
    • 2
  1. 1.Laboratoire Sciences de l’Information et des Systèmes-LSIS CNRS UMR6168 
  2. 2.Département de BiologieUniversité du Sud Toulon-VarLa GardeFrance

Personalised recommendations