摘要
本文将原连续时间电压域点格神经网络（CNN）模型[1]映射到离散时间电流域，并采用开关电流（SI）技术来进行离散时间CNN的VLSI设计。给出了基本积木块的实现电路，并进行了连通片检测方面的简单应用研究。计算机仿真与理论结果相吻合，所得电路非常适合于用标准CMOS工艺集成，具有潜在的应用前景。
The VLSI implementation of discrete-time cellular neural networks(CNNs) using switched-current(SI) techniques is presented in this paper by mapping the original CNN model[1] from the continuous-time voltage domain into the discrete-time current one The circuit realisation of basic building blocks is illustrated. And a simple application of the CNN in the connected component detection is discussed. Computer simulations agree well with the theory. The resulting circuits are very suitable for integration with standard CMOS technology and will have potential applications.
出处
《电路与系统学报》
CSCD
1997年第1期36-39,共4页
Journal of Circuits and Systems
基金
中国博士后科学基金
关键词
点格神经网络
开关电流
电路理论
信号处理
Cellular neural networks,Switched-current, Circuit theory and design,Signal processing
