基于梯度算子的蚁群图像分割算法研究  

Research on image segmentation by ant colony algorithm based on gratitude operator

作　　者：薛琴[1] 陈玮[1] 罗俊奇[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510090

出　　处：《计算机工程与设计》2007年第23期5660-5663,共4页Computer Engineering and Design

基　　金：广东省教育厅自然科学基金项目(z03032)

摘　　要：提出了一种基于梯度算子的改进蚁群图像分割算法,解决了用传统分割方法很难将目标与背景灰度值相似图割的难题。该算法基于经典的梯度算子图像分割,从聚类的角度出发,综合像素的灰度、梯度特征进行特征分割。蚁群是一种具有离散性、并行性、鲁棒性和模糊聚类能力的进化方法,通过设置不同的蚁群、聚类中心、启发式引导函数和激素来解决蚁群算法循环次数多,计算量大的模糊聚类问题。实验证明,该改进蚁群算法可以快速准确的分割出背景标灰度值极其相似图片的目标图像,是一种有效的图像分割方法。A method of image segmentation by ant colony algorithm based on gratitude operator is presented, The method considers the gray value and gratitude synthetically and introduces into image segmentation from classical threshold of gratitude operator and of clustering. The ant colony algorithm （ACA） is a kind of discrete, parallel, robust and fuzzy clustering ability evolutionary algorithm.Aimed at reducing the time of circulations and shortcoming of mass calculating of an ant colony algorithm, different ant colony, precise the center of cluster, a heuristic function and pheromone are modified for application. The experiment results shows that the improved ACA can detect goal image＇s edge from background fast and well, which is similar the goal image in the gray value, this prove the method is an effective method in image segmentation.

关 键 词：梯度算子 蚁群算法 图像分割 模糊聚类 特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]