一种用于图像二值化的细胞神经网络模型

A Cellular Neural Network Model for Image Binarization

本文提出了一种用于图像二值化的神经网络模型，我们称它为数字式细胞神经网络。它基于蔡少堂等提出的细胞神经网络概念［１］［２］，但采用了神经元离散时间的数字动力学技术。利用这个模型，只要通过简单的整数运算就可以并行高速地对灰度图像进行二值化。对于不同邻接和权值的选择得到了比传统二值化方法更自然的二值图像。另外，所提出的神经网络模型非常适合于数字ＶＬＳＩ实现。

In this paper,we present a neural network model for image binarization,we call the model digital cellular neural network.The model we proposed is based on the concept of cellular neural network, but it uses the digital dynamics of a neuron in discrete times.Through the use of this model,we can do image binarization parallelly and quickly only by doing some simple integer operations.For different neighborhoods and weights,we get more natural binary image than the traditional methods.Moreover,the model can appropriately be implemented using digital VLSI.