摘要
为准确预测农村地区公路网中短期发展规模,研究提出了基于高分遥感影像与地形、区域功能及周边环境划分的农村地区公路网中短期发展规模预测模型。模型采用基于地形区域分类的方法,以深度神经网络提取的路网高分遥感影像为基础数据,综合考虑路网分布、自然村分布及路网提取算法精度等客观情况,并对遥感影像质量偏低导致的识别率下降进行了里程修正。不同地区、不同地理区域下的应用实验结果表明,模型预测结果与当地交通运输主管部门基于项目建设计划汇总的发展目标相比,其拟合度在以山地为主的区域达到95.5%,以平原耕地为主的区域达到94.2%。以上结果表明,预测模型在农村地区公路网中短期规模预测方面具有较好的实用性。
In order to accurately predict the medium and short term development scale of road network in rural areas,this paper put forward a medium and short term development scale prediction model of rural road network based on HR-RS(High-Resolution Remote Sensing)images,division of terrain,area func-tion,and surrounding environment.Using a method based on terrain environment classification,and taking HR-RS images of the road network extracted by deep neuron network as basic data,the model compre-hensively considered the objective conditions such as road network distribution,natural village distri-bution and the accuracy of road network extraction algorithm,and made a mileage correction for the decline of extraction precision caused by the low quality of HR-RS images.The experiment results of the model in different regions and geographical environments showed that:the fitting degree between the prediction results of the model and the development objectives summarized by the local transporta-tion authorities based on the project construction plan in mountainous area was 95.5%,and 94.2%in plain area.It meant that the model was designed reasonably with high practicability in medium and short term road network scale prediction in rural area.
作者
范文涛
马骁
崔应寿
周舟
张淑珍
昌宏哲
FAN Wen-tao;MA Xiao;CUI Ying-shou;ZHOU Zhou;ZHANG Shu-zhen;CHANG Hong-zhe(Digital Transportation Laboratory,China Academy of Transportation Sciences,Beijing 100029,China;Henan Transportation Development Center,Zhengzhou 450000,China)
出处
《交通运输研究》
2022年第1期12-18,共7页
Transport Research
关键词
公路网规模预测
深度神经网络
路网提取
高分辨率遥感影像
公路建设规划
road network scale prediction
deep neural network
road network extraction
high resolution remote sensing image
road construction planning
