基于蚁群聚类的路面识别分析方法  

Road roughness classification method based on ant colony clustering algorithm

作　　者：刘庆华[1] 翟刘佳 江燕燕[1] 张利敏[1] 

机构地区：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003

出　　处：《江苏科技大学学报（自然科学版）》2014年第5期496-499,504,共5页Journal of Jiangsu University of Science and Technology:Natural Science Edition

基　　金：国家自然科学基金资助项目(51008143)

摘　　要：道路是城市的重要基础设施.路面状况的鉴定与评价直接关系到公路的养护决策,是一项至关重要的工作.本研究将蚁群聚类算法应用于路面识别领域,提出一种基于蚁群聚类优化的路面识别分析方法,将此方法运用于路面识别系统.该系统由数据采集、数据预处理和蚁群聚类优化3个模块组成,其中蚁群聚类主要提取了纵轴方向的3个特征属性,通过不断更新聚类中心产生的信息素矩阵,并通过分类器进行路面分类.实验结果表明:蚁群聚类优化具有全局收敛以及启发式学习等优点,运算效率高,路面识别准确率达到95.3%,验证了此方法的有效性.Roads are one of the most important infrastructures in a city and the identification and evaluation of their pavement conditions is a vital task directly related to roadway maintenance.In this paper, a new road roughness analysis algorithm of ant colony clustering is applied to the road recognition.The system consists of da-ta acquisition, data preprocessing and ant clustering optimization modules among which the ant colony clustering method mainly extracts three attributes of the longitudinal direction and update the center of cluster according to the continuously generated pheromone matrix and classify pavement classifier.Experiments show that this algo-rithm has high operation efficiency due to global convergence of the ant colony optimization and heuristic learn-ing, and the road recognition accuracy rate is up to 95.3%.

关 键 词：蚁群聚类 信息素矩阵 路面分级 特征提取 

分 类 号：U416.217[交通运输工程—道路与铁道工程]