Optimization of the Hölder Image Descriptor using a Genetic Algorithm

Published in Conferences Papers
  1. Leonardo Trujillo, Pierrick Legrand, Gustavo Olague and Cynthia B Pérez. Optimization of the hölder image descriptor using a genetic algorithm. In GECCO. 2010, 1147-1154. BibTeX

    	author = "Leonardo Trujillo and Pierrick Legrand and Gustavo Olague and Cynthia B. P{\'e}rez",
    	title = {Optimization of the h{\"o}lder image descriptor using a genetic algorithm},
    	booktitle = "GECCO",
    	year = 2010,
    	pages = "1147-1154",
    	ee = "http://doi.acm.org/10.1145/1830483.1830695",
    	crossref = "DBLP:conf/gecco/2010",
    	bibsource = "DBLP, http://dblp.uni-trier.de"
  2. Martin Pelikan and Jürgen Branke (eds.). Genetic and Evolutionary Computation Conference, GECCO 2010, Proceedings, Portland, Oregon, USA, July 7-11, 2010. ACM, 2010. BibTeX

    	editor = {Martin Pelikan and J{\"u}rgen Branke},
    	title = "Genetic and Evolutionary Computation Conference, GECCO 2010, Proceedings, Portland, Oregon, USA, July 7-11, 2010",
    	booktitle = "GECCO",
    	publisher = "ACM",
    	year = 2010,
    	isbn = "978-1-4503-0072-5",
    	bibsource = "DBLP, http://dblp.uni-trier.de"

Local image features can provide the basis for robust and invariant recognition of objects and scenes. Therefore, compact and distinctive representations of local shape and appearance has become invaluable in modern computer vision. In this work, we study a local descriptor based on the Hölder exponent, a measure of signal regularity. The proposal is to find an optimal number of dimensions for the descriptor using a genetic algorithm (GA). To guide the GA search, fitness is computed based on the performance of the descriptor when applied to standard region matching problems. This criterion is quantified using the F-Measure, derived from recall and precision analysis. Results show that it is possible to reduce the size of the canonical Hölder descriptor without degrading the quality of its performance. In fact, the best descriptor found through the GA search is nearly 70% smaller and achieves similar performance on standard tests.

Published in
GECCO '10 Proceedings of the 12th annual conference on Genetic and evolutionary computation
Pages 1147-1154
Date of conference
07 - 11 July 2010