Advanced search
Start date

Challenges in multilevel design of morphological operators


Designing morphological operators with good performance in problems involving image processing and analysis is not a trivial task. A useful approach to help the design process is its formulation as a machine learning problem: pairs of input-output images are used as training sample to generate an operator that maps input images into their respective output images. In our investigation context, we consider operators that depend on a neighborhood around a point to be processed. Small neighborhood imposes a strong constraint resulting in constraint error, whereas a large window results in large variance due to statistical imprecision. For binary image operators, a multilevel design approach has been recently proposed as a way to deal with the trade-off between these two types of errors. Experimental results indicate that the approach is promising. In the multilevel approach, outcomes of the previous levels are combined at each new training level. The choice of parameters in the multilevel approach is so far done manually. In this project we propose to further investigate practical and theoretical aspects of the multilevel approach; in particular, we are interested on ways to automate the choice of parameters and on extending the approach to gray-scale image operators. (AU)

Articles published in Agência FAPESP Newsletter about the research grant:
Articles published in other media outlets (0 total):
More itemsLess items

Scientific publications
(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)
DORNELLES, MARTA M.; HIRATA, NINA S. T.; IEEE. A genetic algorithm based approach for combining binary image operators. 2012 21ST INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR 2012), v. N/A, p. 4-pg., . (11/00325-1)
MONTAGNER, IGOR S.; HIRATA, NINA S. T.; HIRATA, JR., ROBERTO. Staff removal using image operator learning. PATTERN RECOGNITION, v. 63, p. 310-320, . (11/23310-0, 15/17741-9, 11/00325-1, 14/21692-0)

Please report errors in scientific publications list using this form.