摘要
以量子细胞自动机为神经元,以量子细胞自动机的极化率和量子相位为状态变量,提出了一种二维多层的量子细胞神经网络结构;以量子细胞自动机的极化率为像素值,通过选择不同的模板,使得量子细胞神经网络实现水平线和垂直线检测以及图像去噪等功能,仿真结果显示其在图像处理上的有效性。与传统的CNN相比,量子细胞神经网络易于超大规模实现且具有超低功耗、超高集成度等优点。
The two-dimensional large-neighborhood quantum cellular neural network（CNN） is proposed by using the quantum cellular automata as neuron cells, with the polarization of a quantum-dot cell and quantum phase as state variables. The proposed quantum cellular neural network can perform the functions of horizontal line detecting, vertical line detecting and image noise removal by using the polarization of quantum-dot cell as pixel value and selecting different cloning templates. The simulation results demonstrate the effectiveness of the proposed quantum cellular neural network. Moreover, compared with the traditional CNN, the quantum cellular neural network possesses the advantages of the simplified interconnection, the extremely high packing densities and the extremely low power consumption.
出处
《固体电子学研究与进展》
CAS
CSCD
北大核心
2008年第3期340-345,共6页
Research & Progress of SSE
基金
陕西省自然科学基础研究计划项目(2005F20)
空军工程大学理学院学位论文创新基金(2007B003)
关键词
量子细胞神经网络
量子细胞自动机
细胞神经网络
图像处理
quantum cellular neural network
quantum cellular automata
cellular neural network
image processing
