| 轨迹延续性与影像特征相似性结合的城市道路提取 |
| |
| 引用本文： | 方志祥,仲浩宇,邹欣妍.轨迹延续性与影像特征相似性结合的城市道路提取[J].测绘学报,1957,49(12):1554-1563. |
| |
| 作者姓名： | 方志祥 仲浩宇 邹欣妍 |
| |
| 作者单位： | 武汉大学测绘遥感信息工程国家重点实验室, 湖北 武汉 430079 |
| |
| 基金项目： | 国家重点研发计划（2017YFB0503802）；国家自然科学基金面上项目（41771473）；中央高校基本科研业务费专项（2042020kfxg24） |
| |
| 摘 要： | 城市道路区域检测是城市土地管理、交通规划等领域的迫切需求，而传统城市道路区域检测多使用轨迹提取、遥感解译、人工采集等单独方式，在自动化程度或提取质量上存在一定的局限性。本文结合GNSS轨迹点与高分遥感影像各自的数据优势，提出一种基于轨迹延续性与影像特征相似性的遥感影像道路区域检测方法。该方法以出租车GNSS轨迹点构建轨迹特征栅格，基于轨迹延续性在平均方向特征栅格中划分路段对象，利用道路对象的光谱特征向轨迹无法覆盖的小区内部进行拓展，以获得提取区域内较为完整的道路信息。试验证明：本文方法可以有效降低道路的同物异谱现象及阴影、树木遮挡的影响，高效地提取高分遥感影像中的道路区域。与传统的遥感影像分类方法相比，具有更高的精度与自动化程度，相较于深度学习模型具有更广的适应性。
|
| 关 键 词： | 高分遥感影像 出租车轨迹 轨迹连续性 影像特征相似性 道路区域检测 |
| 收稿时间： | 2019-09-24 |
| 修稿时间： | 2020-08-12 |
Extracting urban road area based on combination of trajectory continuity and image feature similarity |
| |
| FANG Zhixiang,ZHONG Haoyu,ZOU Xinyan.Extracting urban road area based on combination of trajectory continuity and image feature similarity[J].Acta Geodaetica et Cartographica Sinica,1957,49(12):1554-1563. |
| |
| Authors: | FANG Zhixiang ZHONG Haoyu ZOU Xinyan |
| |
| Institution: | State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China |
| |
| Abstract: | Urban road area detecting is the imperious demand in the area of management of city land use, transportation planning and so on. Trajectory extraction, remote sensing image classification and artificial collection are the traditional methods for road network detection with some limits on automation degree or extraction quality. This paper proposes a method for detecting road area in high-resolution remote sensing image based on trajectory continuity and image feature similarity, and this method utilizes the advantages of GNSS trajectory and remote sensing image. The proposed methods could be divided into three steps: firstly, using GNSS trajectory points to construct images of trajectory feature and selecting the high-confidence grids with high density value. Secondly, generating road objects based on trajectory continuity in average direction feature image. Thirdly, dividing high-resolution remote sensing image into several small areas by using road segments and extending road areas based on image feature similarity automatically to detect roads which not covered by trajectory. The experiment evidences that this method could detect road areas efficiency and accuracy in high-resolution remote sensing image and decreasing the bad effect on the different roads with different spectrums. Compared with the traditional remote sensing image classification methods, the proposed method has a higher precision and automatic degree. |
| |
| Keywords: | |
|
| 点击此处可从《测绘学报》浏览原始摘要信息 |
| 点击此处可从《测绘学报》下载免费的PDF全文 |
