Fish age classification based on length, weight, sex and otolith morphological features

Knowledge of age in fish populations is crucial in stock assessment and management. Currently, some pattern recognition systems has been proposed for accomplishing automatically this task based on extracting different kind of features from fish otholiths as well as other features related to fish. However, there is no clear evidence on which features are best for age classification. In this work, we compare otholith morphological features versus other features like fish length, weight and sex. The accuracy has been tested for different support vector machine classifiers using a cod database. As demonstrated, fish length, weight and sex are slightly superior to otholith morphological features for age classification purposes. However, it is the synergistic combination of both kinds of features that achieves the greatest accuracy (∼75%).