摘要
在文献[1]的基础上,提出了一个基于动态核的形态联想记忆网络方法,特点是同一幅图像,如果其所含的噪声情况不同,则其核也将不同,从而较好地解决了图像含有随机噪声时的联想记忆问题。实验证明,此方法具有良好的性能,双向联想记忆的准确率优于文献[1]中介绍的方法。
The new morphological associative memory based on dynamic kernel is presented. The property of the new method is that the kernel of the same image varies with different noise on it. In the method, the question of associative memories for images with random noise is solved well. The accuracy rate is superior to the method, which is verified by the experiments.
出处
《计算机工程与设计》
CSCD
北大核心
2005年第11期2907-2909,共3页
Computer Engineering and Design
基金
江苏省自然科学基金项目(BK2003017)
关键词
联想记忆
形态学
形态学神经网络
核动态核
associative memories
morphology
morphological neural networks kernel
dynamic kernel.
