一种基于蚁群算法的边缘提取算法  

An Edge Extraction Method Based on Gradient with ACS

作　　者：陈亮[1] 郭雷[1] 

机构地区：[1]西北工业大学自动化学院,西安710072

出　　处：《光子学报》2010年第4期759-763,共5页Acta Photonica Sinica

基　　金：国家自然科学基金(60675015)资助

摘　　要：提出一种基于梯度定义和蚁群算法的边缘提取算法.根据梯度在边缘处的直观响应,提出自适应的边缘特征定义,并结合蚁群算法的特性,令蚂蚁遗留不同等级的信息素.根据搜索权值矩阵得到尽量平滑的边缘,最终的边缘通过自动阈值进行提取.实验表明:该算法的性能在同样的智能算法和经典的边缘检测算子中表现良好,并对噪音不敏感.A new edge extraction method based on ACS and gradient is proposed. Due to intormat,on feedback, the complicated problem can be resolved by the simple ants. The new food definition and movement meehanism to the ant is applied in this study. Multi-scale gradient is taken as the food objective of the ant. Besides, two directional functions included strong and weak one for prior directional searching are used, and the pheromone is graded. Furthermore, the thresholding is automatic for the final edge extraction. According to the compare with other methods in the experiment,the efficieney of the proposed method is proved, and the method is not sensitive to the noise.

关 键 词：多尺度梯度 蚁群算法 边缘提取 自适应 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]