Image segmentation with the growing neural gas

Abstract

Over many years algorithms have been proposed as methods for the segmentation of images. Everytime the work has been improved and optimized for better results with much faster algorithms and newer ways to adapt the algorithm to the needs of real world requirements. In this article we use the Growing Neural Gas on RGB images and carry out various forms of experimentations to analyze the segmentation of RGB images by this novel algorithm. There are two phases to the segmentation process where in the rst phase the input data is learnt by the neural network to produce a model, and later the second phase includes the exploitation of this model to produce the segmented images. Experimentations include changing the factors that a ect the algorithm and also the segmentation process.