摘要
针对单一聚类算法在图像分割中容易陷入局部最优或有过分割现象,造成分割精确度低等问题,文章提出了基于K-均值聚类和蚁群聚类相结合的新算法。新算法先将K-均值算法作快速分类,根据K-均值分类结果更新蚂蚁各路径上的信息素,指导其他蚂蚁选择,以提高蚁群聚类算法的运行效率。实验结果证明,新算法在图像分割处理的精确度上较单一的K均值和蚁群聚类算法有很大提高。所以进一步表明该方法对于图像分割具有很好的通用性和有效性,是一种实用的、有前途的图像分割方法。
For a single clustering algorithm for image segmentation is easy to fall into local optimum or the phenomenon of over-segmentation,which will result in low accuracy problem of segmentation,We put forward a new clustering algorithm combining ant colony clustering with K-means algorithm.The new clustering algorithm is the first K-means algorithm for rapid classification,and according to classification results update the pheromones to guide the choosen of other ants,to improve the efficiency of ant colony clustering algorithm.Experimental results show that the new algorithm for image segmentation accuracy than a single K means clustering algorithm and the ant colony clustering algorithm has greatly improved.Therefore,further show that this method for image segmentation has good versatility and effectiveness,is a practical and promising method of image segmentation.
出处
《计算机与数字工程》
2011年第6期138-141,共4页
Computer & Digital Engineering
基金
江苏省高校自然科学基础研究项目(编号:08KJB520003)资助
关键词
蚁群聚类
K-均值聚类
图像分割
ant colony clustering
K-means clustering
image segmentation
