一种图像分割的人工神经网络新方法

A New Neural Network Method for Image Segmentation

本文着重研究应用人工神经网络（ＡＮＮ）理论来进行图像分割的问题，提出了一种图像分割的ＡＮＮ新模型，详细地研究了该模型的动力学特性，并给出了该动力学特性的计算机实现算法。利用实拍图像对此ＡＮＮ模型进行了一些有意义的仿真实验，并与常用的矩保持门限分割方法作比较，结果表明：本文提出的图像分割ＡＮＮ模型是切实可行的，且能较好地完成低信噪比（ＳＮＲ）条件下图像的分割任务，其分割结果优于矩保持门限分割方法。

In this paper,the problem of image segmentation using artificial neural network(ANN) is discussed.A new ANN model for image segmentation is proposed. The dynamics of the proposed model is described in details and the corresponding alogrithm realized with computer is obtained. Some simulations are run using both the proposed model and the moment-preserving thresholding. respectively,and the corresponding results are obtained.