Image segmentation by neural-net classifiers with genetic selection of feature indices
K. Matsui, Y. Kosugi · 2003
A new method of the image segmentation using neural-net classifiers and GA-based feature selection is proposed. We use the genetic algorithm (GA) to select the optimal combinations of feature indices as the input of the neural-net classifiers. Our GA method of feature selection is based on the new evaluation function, VQCCE. By this measurement, we can evaluate the combination of feature indices rapidly without testing on the actual classifiers. We use two types of neural-net classifiers: the backpropagation network and a modified counter-propagation network. We applied our method to some classification problems and showed the effectiveness of our method.