Segmentation of Color Images Using Multi-FART Network

This paper proposed a multi-FART neural network for color image segmentation. The network received three transformed color components - hue, value and saturation. Then, using multi-FART network categorical ability, its output will send the categorized values to the decision layer. After combined postprocessing, the right categorization of color vector is output from the last layer. Compared with color watershed algorithm, experiments on color image show promising results.