A New Unsupervised Approach to Face Recognition
A new unsupervised approach to face recognition is proposed in this paper. Shape and color entropy is presented to descript face features. Firstly, images are pre-processed including face normalization and image segmentation and so on. Secondly, by using the information entropy theory, the method defines the color and shape entropy of the face images, respectively. Finally, an integrated similarity measurement framework is presented by computing mutual information between images according to these entropies. Compared with other methods of feature description, experiments indicate that this approach is more effective and efficient.
Keywordsmutual information face recognition image similarity feature description
Unable to display preview. Download preview PDF.
- 5.Liu, C., Wechsler, H.: A unified Bayesian framework for face recognition. In: Proc. Internat. Conf. on Image Processing (ICIP 1998), pp. 151–155 (1998)Google Scholar
- 11.Viola, P., Wells, W.: Alignment by maximization of mutual information. In: Proceedings of the 5th International Conference on Computer Vision, Boston, MA, pp. 16–23 (1995)Google Scholar
- 12.Collignon, A., Maes, F., Vandermeulen, D., et al.: Automated multimodality image registration using information theory. In: Proceedings of the Information Processing in Medical Imaging Conference, Dordrecht, pp. 263–274 (1995)Google Scholar
- 13.Fan, Z.Z., Zhou, S.C.: Image Retrieval Based on Shape Entropy. Journal of Computer Application & Research 24(9), 309–311 (2007) (in Chinese)Google Scholar
- 14.ORL face database (2008), http://www.uk.research.att.com/facedatabase.html
- 15.Yale face database (2007), http://cvc.yale.edu/projects/yalefaces/yalefaces.html