湖北农业科学 ›› 2025, Vol. 64 ›› Issue (10): 190-194.doi: 10.14088/j.cnki.issn0439-8114.2025.10.029

• 信息工程 • 上一篇    下一篇

面向类城市路网空间的多时相多地形耕地地块边界提取研究

郑明雪, 沈祥成, 罗治情, 陈娉婷, 官波, 马海荣   

  1. 湖北省农业科学院农业经济技术研究所/湖北省农业科技创新中心农业经济技术研究分中心/湖北省乡村振兴研究院,武汉 430064
  • 收稿日期:2025-02-08 出版日期:2025-10-25 发布日期:2025-11-14
  • 通讯作者: 马海荣(1986-),女,山东菏泽人,助理研究员,博士,主要从事遥感技术农业应用研究,(电话)15002778923(电子信箱) mahairong1008@126.com。
  • 作者简介:郑明雪(1989-),女,湖北洪湖人,助理研究员,博士,主要从事遥感技术农业应用研究,(电话)15171503311(电子信箱)zhengmingxue1128@163.com。
  • 基金资助:
    国家自然科学基金青年科学基金项目(NSFC42201397); 湖北省自然科学基金一般面上项目(2024AFC048); 湖北省农业科技创新中心“农业经济与信息研究”团队项目(2025-620-000-001-025)

Research on farmland parcel boundary extraction oriented to urban road network-like space using multi-temporal and multi-terrain images

ZHENG Ming-xue, SHEN Xiang-cheng, LUO Zhi-qing, CHEN Ping-ting, GUAN Bo, MA Hai-rong   

  1. Institute of Agricultural Economy and Technology, Hubei Academy of Agricultural Sciences/Hubei Agricultural Science and Technology Innovation Center Agricultural Economic and Technological Research Sub-Center/Hubei Rural Revitalization Research Institute, Wuhan 430064, China
  • Received:2025-02-08 Published:2025-10-25 Online:2025-11-14

摘要: 通过挖掘耕地地块边界与城市路网在空间分布上的内在关联,将不同区域的耕地地块边界分布类比为不同等级的道路分布,构建耕地地块边界的类城市路网空间。选取湖北省洪湖市和通山县作为研究区域,采用无人机遥感影像作为数据源,通过多尺度方法提取多时相多地形影像的颜色、亮度与纹理特征,进而实现耕地地块边界的精确识别。结果表明,该方法在不同季节与地形条件下对耕地地块边界提取的效果具有显著差异。对于影像色彩丰富、地块规整的区域,该方法能够完整提取边界,且多尺度参数可有效满足不同细节层次的需求。在休耕期或丘陵地区等场景下,受杂草、水体、浮萍及裸露土壤等多变地表覆盖的干扰,地块特征可分性下降,说明该方法在特征复杂的场景中仍存在局限性。

关键词: 类城市路网空间, 遥感影像, 多时相, 多地形, 耕地, 地块边界提取

Abstract: By exploring the intrinsic spatial distribution relationship between farmland parcel boundaries and urban road networks, the distribution of farmland parcel boundaries in different regions was analogized to the distribution of different road grades, constructing an urban road network-like space for farmland parcel boundaries.Honghu City and Tongshan County in Hubei Province were selected as study areas. Using UAV remote sensing images as data sources, color, brightness, and texture features of multi-temporal and multi-terrain images were extracted through multi-scale methods, thereby achieving accurate identification of farmland parcel boundaries.The results showed that the effectiveness of this method for farmland parcel boundary extraction varied significantly across different seasons and terrain conditions.In areas with rich image colors and regular parcels, the method could completely extract boundaries, and multi-scale parameters effectively met the requirements at different levels of detail.In scenarios such as fallow periods or hilly areas, the separability of parcel characteristics decreased due to interference from diverse surface covers like weeds, water, duckweed, and bare soil, indicating that the method still had limitations in feature-complex scenarios.

Key words: urban road network-like space, remote sensing image, multi-temporal, multi-terrain, farmland, parcel boundary extraction

中图分类号: