基于自组织特征映射网络SOM的图像分割方法

Image-Segmentation based on SOM neural network

阈值法是图像分割中常用的有效分割技术之一,但是阈值的选取对分割结果影响很大,特别是对于生物医学图像,分割结果往往不能满足要求.本文将SOM神经网络应用到图像分割中,利用其自组织学习的能力自动获取阈值,其结果不仅优于传统阈值法,而且也优于竞争型神经网络.

Threshold method is commonly used in image segmentation as a effective segmentation technique,but the selection of threshold has a great impact on the results,especially for biomedical images,the segmentation results often do not match the requirements.This paper used SOM neural network applications to image segmentation,uses of its self-organizing learning capabilities to Automatic Generate threshold,the result is not only superior to the traditional threshold,but also superior to the competitive neural networks.