基于航空影像提取交通路网设计与实现

Design and implementation of traffic network extraction based on aerial images

交通对国民经济发展影响极大,及时有效的更新交通路网信息服务居民出行获得关注。传统人工外业测量采集道路信息的方法成本高效率低;而航空影像具有更新快、时效性强的特点,研究如何从航空影像中提取交通路网具有重大现实意义和经济价值。本文利用最大似然法、迭代自组织聚类算法、最邻近分类和隶属度模糊分类四种算法从航空影像中提取交通路网。前两种基于像元的方法总体精度达0.8以上,后两种面向对象的方法总体精度最高达0.95。实验结果表明基于航空影像提取交通路网可行且具有较高精度,实际工程中可根据影像地物类型合理选择提取算法。

Traffic construction has a great impact on the development of the national economy.Therefore,timely and effective updating of traffic road network information to serve residents’travel has attracted more and more attention.The traditional method of collecting road information by artificial field measurement which cost a lot of money and tim,while aerial imagery has the characteristics of fast updating and strong timeliness.So it is of great practical significance and economic value to study how to extract traffic road network from aerial imagery.In this paper,four algorithms including maximum likelihood method,ISODATA,nearest neighbor classification and membership fuzzy classification are used to extract traffic network from aerial images.The overall accuracy of the first two pixel-based methods is over 0.8,and the overall accuracy of another object-oriented methods is over 0.9.The experimental results show that the extraction of traffic road network based on aerial images is feasible and has high accuracy.In practical projects,the extraction algorithm should be reasonably selected according to the type of image features.