摘要
给出了一种基于自组织聚类神经网络的图像分割算法．针对聚类中心初始值选取的盲目性，提出了初始值优选法，大幅度提高了分割算法的速度．实验表明，文中提出的算法能快速、准确地分割医学图像，将原始照片中不易分辨的病灶清晰地呈现出来．
An amended clustering neural network algorithm is presented by applying the Kohonen clustering neural network in medical image segmentation. Also, the “optimum seeking method of initial value” is presented, which overcomes the default of traditional clustering methodologies and greatly increases the speed. The experiments show that medical pictures can be segregated quickly and correctly with the new method, and the blood lump or the focus of infection which is not easy to distinguish in an initial image, can be seen clearly.
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
1998年第5期602-605,共4页
Journal of Xidian University
关键词
神经网络
图像分割
医学图像处理
聚类分析
neural network image segmentation medical image processing clustering methodology quicker method
