摘要
利用PBM模糊聚类有效性函数以图像特征空间为搜索空间,实现有效性函数的全局寻优,用并行小生境技术解决粒子群(PSO)算法的早收敛问题,优化聚类的全局收敛性能,实现有效聚类数目与聚类中心的并行寻优。通过对遥感图像分割的实验证明,与传统粒子优化群算法的分割结果相比,本文算法拥有更高的有效性且分割效果更优。
An image segmentation algorithm is proposed by utilizing ：fuzzy clustering which is based on the PBM validity index. The algorithm makes use of swarm intelligent algorithm for searching optimal validity index in image feature space. The parallel niching technique is used for solving the premature convergence of the original swarm optimization algorithm,and optimizing the clustering global convergence. The optimization of both clustering number and clustering centroids is realized. The simulation for remote sensing image segmentation shows that the proposed algorithm has a better segmental result compared with the canonical swarm intelligent algorithm.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2010年第4期618-622,共5页
Journal of Optoelectronics·Laser
