测绘科学技术学报2025，Vol.41Issue(4)：419-426,8.DOI:10.3969/j.issn.1673-6338.2025.04.012
一种利用轨迹道路匹配特征检测道路网错误的BiLSTM模型
A BiLSTM Model for Detecting Road Network Errors Using Trajectory Road Matching Features
摘要
Abstract
With the continuous development of fields such as travel navigation and autonomous driving,the accuracy and currency of road networks are increasingly required.In order to facilitate the timely identification of issues within the road network,a BiLSTM model is proposed for detecting road network errors using trajectory road matching features in this paper.The study consists of two principal components,feature extraction and model con-struction.Partial map matching algorithm is used to match vehicle trajectories and roads,and on the basis of the features used in algorithm,Zernike moments are introduced to describe the geometric shape of the trajectories,and then the trajectory point feature vectors are constructed.A BiLSTM model is then designed to classify the trajectory points using the trajectory point feature vectors,so as to detect the faulty road sections in the road network.Experi-ments show that the addition of Zernike moments helps to improve the classification performance of the deep learn-ing model,and the proposed BiLSTM model outperforms the traditional machine learning methods and the original bidirectional recurrent neural network(BiRNN)model in many evaluation indices.关键词
车辆轨迹数据/城市道路网/轨迹道路匹配/道路网错误检测/深度学习Key words
vehicle trajectory data/urban road network/trajectory road matching/road network errors detection/deep learning分类
测绘与仪器引用本文复制引用
孙士杰,孙群,张付兵,李少梅,陆川伟..一种利用轨迹道路匹配特征检测道路网错误的BiLSTM模型[J].测绘科学技术学报,2025,41(4):419-426,8.基金项目
国家自然科学基金项目(42101454 （）
42101455). （）
