摘要
图像分割是图像分析与理解的一个关键课题.本文提出利用两个竞争Hopfield神经网络(CHNN)实现灰度图像的自动聚类分割.其中第一个CHNN根据灰度分布自动确定聚类数目并实现预分割,第二个CHNN在预分割基础上融入邻域相关信息实现最终的图像分割.理论和实验表明:该方法不仅能自动确定聚类数,而且具有收敛速度快、抗噪能力强等优点.
Image segmentation is the most important problem in image analysis and understanding. An approach using two competitive Hopfield neural networks (CHNN) is proposed for automatic image segmentation by clustering. The first CHNN is used to determine the clustering number automatically and segment the image based on the gray level distribution. The second CHNN is used to incorporate the neighboring information for image segmentation based on the result, of the first CHNN. Experimental results show that, the approach not, only can determine the clustering number automatically, but also has the advantage of fast, convergent speed, insensitiveness to noise, etc.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1998年第2期215-221,共7页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金
