贾时银, 周冬明, 聂仁灿, 赵东风. 脉冲耦合神经网络模型参数优化及图像分割[J]. 云南大学学报(自然科学版), 2010, 32(5): 521-525 .
引用本文: 贾时银, 周冬明, 聂仁灿, 赵东风. 脉冲耦合神经网络模型参数优化及图像分割[J]. 云南大学学报(自然科学版), 2010, 32(5): 521-525 .
JIA Shi-yin, ZHOU Dong-ming, NIE Ren-can, ZHAO Dong-feng. Pulse coupled neural network model parameter optimizationand image segmentation[J]. Journal of Yunnan University: Natural Sciences Edition, 2010, 32(5): 521-525 .
Citation: JIA Shi-yin, ZHOU Dong-ming, NIE Ren-can, ZHAO Dong-feng. Pulse coupled neural network model parameter optimizationand image segmentation[J]. Journal of Yunnan University: Natural Sciences Edition, 2010, 32(5): 521-525 .

脉冲耦合神经网络模型参数优化及图像分割

Pulse coupled neural network model parameter optimizationand image segmentation

  • 摘要: 脉冲耦合神经网络在图像处理中有着重要应用,但存在模型参数难以选择和图像边缘过于平滑问题.通过对图像进行双线性插值运算,再利用具有保护图像边缘作用的各向异性扩散特性确定模型的链接权值参数,采用遗传算法求解模型的链接强度参数和衰减阈值参数,成功实现了图像的自动分割.仿真结果表明,该方法得到的图像分割结果,体现了更多的图像轮廓和边缘细节,具有较好计算性能.

     

    Abstract: Pulse coupled neural networks in image processing has important applications,but there are difficulties in choosing model parameters and image edges are too smooth problems.In this paper,the image bilinear interpolation operator is performed,and the anisotropic diffusion characteristics in images are used to determine the link weight parameter of the model.By using genetic algorithms to solve the model parameters and the attenuation of the link strength threshold parameter,the automatic image segmentation is successfully implemented.The simulation results show the image segmentation by the algorithm embodies more image contour and edge details,so the algorithm has good computing performance.

     

/

返回文章
返回