Automatic Multilevel Thresholding for Image Segmentation by the Growing Time Adaptive Self-Organizing Map.

Hamed Shah-Hosseini, Reza Safabakhsh

Abstract: In this paper, a Growing TASOM (Time Adaptive Self-Organizing Map) network called "GTASOM" along with a peak finding process is proposed for automatic multilevel thresholding. The proposed GTASOM is tested for image segmentation. Experimental results demonstrate that the GTASOM is a reliable and accurate tool for image segmentation and its results outperform other thresholding methods.