一种基于链码的农村公路线形比对与筛选算法  

A Rural Highway Alignment Comparison and Screening Algorithm Based on Chain Code

作　　者：范文涛[1] 孙翠羽 崔应寿[1] 刘柳杨[1] 龙佳宁 FAN Wentao;SUN Cuiyu;CUI Yingshou;LIU Liuyang;LONG Jianing(China Academy of Transportation Sciences,Beijing 100029,China;School of Transportation Engineering,East China Jiaotong University,Nanchang 330013,China;Applied Translation Research Center of Transportation and Engineering,East China Jiaotong University,Nanchang 330013,China)

机构地区：[1]交通运输部科学研究院,北京100029 [2]华东交通大学交通运输工程学院,江西南昌330013 [3]华东交通大学交通与工程应用翻译研究中心,江西南昌330013

出　　处：《交通运输研究》2025年第1期102-110,共9页Transport Research

基　　金：国家重点研发计划项目(2021YFE0105600);国家自然科学基金面上项目(51978263)。

摘　　要：为提升农村公路电子地图数据校核的效率和精确性,提出一种基于链码技术的农村公路线形比对与筛选算法模型。首先,引入链码技术对农村公路电子地图中的路线进行识别和提取,获得路线的起点、拐点及方向变化等线形特征信息,并实现不同地理区域和时间维度下的路线线形链码特征信息的精确匹配。其次,选取我国东部、中部、西部区域中典型区县的2022及2023年农村公路线形数据,对算法模型进行训练,得出路线线形链码差异的不同参数阈值;对于超出阈值的农村公路路线数据,再运用卷积神经网络算法从遥感影像中提取路线线形并获取线形链码特征信息进行比对筛选以及影像评价。最后,将该算法应用于2023年度全国农村公路电子地图中459.9万km的路线数据进行解算验证。结果表明,与传统“全重叠”方法相比,该算法效率提升了72.1%,识别率从64.5%提升至90.6%,准确率从95.7%提升到97.3%。研究证明,该算法显著提高了电子地图线形数据处理的效率和准确率,可为农村公路基础数据入库提供技术支撑,提升农村公路数字化发展水平。To improve the efficiency and accuracy of rural highway electronic map data verification,a rural highway alignment comparison and screening algorithm model based on chain code technology was proposed.Firstly,the chain code technology was introduced to identify and extract the routes in the electronic map of rural highway,the linear feature information such as the starting point,inflection point and direction change of the route was obtained,and the accurate matching of the linear chain code feature information of the route in different geographical regions and time dimensions was realized.Secondly,the 2022 and 2023 rural highway alignment data of typical districts and counties in the eastern,central,and western regions of China were selected to train the algorithm model,and the different parameter thresholds for the difference in route linear chain codes were obtained.For rural highway route data that exceed the thresholds,a convolutional neural network algorithm was used to extract route alignments from remote sensing images and obtain linear chain code feature information for comparison,screening,and image evaluation.Finally,the algorithm was applied to the route data of 4.599 million kilometers in the 2023 national rural highway electronic map for calculation and verification.The results show that compared with the traditional"total overlap"method,the efficiency of the algorithm was increased by 72.1%,the recognition rate was increased from 64.5%to 90.6%,and the accuracy was increased from 95.7%to 97.3%.The research proves that the algorithm significantly improves the efficiency and accuracy of electronic map linear data processing.It can provide technical support for the basic data storage of rural highway,and improve the level of digital development of rural highway.

关 键 词：链码算法 线形比对 农村公路 遥感影像 卷积神经网络 

分 类 号：U495[交通运输工程—交通运输规划与管理]