This paper is dedicated to the problem of two-dimensional shape recognition from the point of view of possible optimization regarding feature extraction and classification methods. Moment invariant and stochastic AR models are considered as feature extraction methods. For classification, the Bayesian parametric and nonparametric classifiers are analyzed, as well as multilayer perceptron applied as a nonparametric classifier. Evaluation of the considered methods is done on the basis of the Bayes error estimates calculated on the corresponding data sets.
Reviewer: Miodrag Mihaljević (Beograd)