作　　者： (车翔玖）; (张孙旻）;

机构地区： 吉林大学计算机科学与技术学院,长春130012

出　　处： 《吉林大学学报（工学版）》 2017年第5期1577-1582,共6页

摘　　要： 针对蚁群算法在提取边缘不连续以及难以搜索到弱边缘的问题提出了改进措施,利用Otsu预处理,并对信息素矩阵和启发式矩阵初始化做了改进,同时加入信息素异步更新策略和参数自适应修改来避免过早陷入停滞从而发现更多弱边缘。通过主观对比和定量分析,本文方法提取了更多的连续边缘以及弱边缘,具有更好的鲁棒性。 An ant colony algorithm with features of high robustness,distributed computing,and positive feedback,is used to solve edge detection problem.In order to prevent extracting discontinuous edges and missing weak edges by using ant colony algorithm,some improving measures are put forward.An Ostsu's method for pre-processing is proposed,the initialization of pheromones matrix and heuristic matrix is improved;meanwhile,pheromones asynchronous update strategy and parameter self-adaptive modification are applied to avoid early stagnancy.Experiments show that the proposed method can extract more continuous edges and weak edges with better robustness.

关 键 词： 计算机应用 边缘检测 蚁群算法 异步更新