A simple learning algorithm for growing self-organizing maps and its application to the skeletonization
H. Sasamura, Toshimichi Saito · 2004
This paper presents a simple learning algorithm for growing self-organization maps (ab. SOMs) and considers its application to the skeletonization. In order to adapt the shape of the input data, the map can have partial tree and loop topology. In the algorithm, the map can grow and the topology can change based on occasional inspection of learning history of each cell and MST. If the control parameters are selected suitable, the algorithm can be applied effectively for skeletonization of Japanese characters.