湖北农业科学 ›› 2023, Vol. 62 ›› Issue (6): 157-162.doi: 10.14088/j.cnki.issn0439-8114.2023.06.029

• 信息工程 • 上一篇    下一篇

基于高分二号卫星影像的基塘提取与形态分析

蒋海涛1, 周晋皓1,2, 李欣茹1, 林景华1,3, 黄少芳1, 刘吼海4, 钟志艺5   

  1. 1.华南农业大学资源环境学院,广州 510642;
    2.广东省土地利用与整治重点实验室,广州 510642;
    3.广州耘趣网络科技有限公司,广州 510699;
    4.广州市华南自然资源科学技术研究院,广州 510642;
    5.广州恒地信息技术有限公司,广州 510635
  • 收稿日期:2022-05-31 出版日期:2023-06-25 发布日期:2023-07-18
  • 通讯作者: 钟志艺(1984-),男,广东龙川人,高级工程师,硕士,研究方向为农业信息与国土整治规划,(电话)18688433441(电子信箱)zzy_1984@126.com。
  • 作者简介:蒋海涛(2002-),男,广东清远人,在读本科生,专业方向为遥感分类方法,(电话)13684900598(电子信箱)j_the_next@163.com。
  • 基金资助:
    国家自然科学基金项目(42001213); 广东省自然科学基金项目(2018A030313201); 广州市科技计划项目(201804020034)

Extraction and pattern analysis of dike-pond based on Gaofen-2 satellite image

JIANG Hai-tao1, ZHOU Jin-hao1,2, LI Xin-ru1, LIN Jing-hua1,3, HUANG Shao-fang1, LIU Hou-hai4, ZHONG Zhi-yi5   

  1. 1. College of Resources and Environment, South China Agricultural University, Guangzhou 510642, China;
    2. Guangdong Provincial Key Laboratory of Land Use and Consolidation, Guangzhou 510642, China;
    3. Guangzhou Yunqu Network Technology Co., Ltd., Guangzhou 510699, China;
    4. South China Academy of Natural Resources Science and Technology, Guangzhou 510610, China;
    5. HeadGIS Information Technology Co.,Ltd., Guangzhou 510635, China
  • Received:2022-05-31 Online:2023-06-25 Published:2023-07-18

摘要: 为掌握粤港澳大湾区基塘的形态特征,以大湾区龙江镇为研究区域,采用面向对象规则分类方法,从高分二号卫星影像中提取基塘,并使用加权聚合邻近度指数度量基塘的形态特征。结果表明,通过样点检验方式和样区检验方式得到提取的总体精度分别为92.25%、80.25%,反映出不同形态基塘的分类精度差异,样区方式更适合检验高空间分辨率影像的分类精度。研究区域有基塘14.06 km2,其中规则塘占58.46%,主要分布在研究区域的中部和北部,紧凑程度较高,便于扩大养殖增加收入;不规则塘占41.54%,主要分布在研究区域的东部和西部,紧凑程度较低,利于种植促进水陆交互。

关键词: 基塘, 面向对象分类, 高分二号, 空间形态

Abstract: For the purpose of analyzing the dike-pond pattern in Guangdong-Hong Kong-Macao Greater Bay Area, the rule-based classification method to extract the dike-pond from GF-2 satellite image was used. Then the dike-pond pattern was measured through Weighted Aggregation and Closeness (WAC) metric. The results showed that the overall accuracy of extraction was 92.25% by sample point test, and 80.25% by sample region test. The sample region test could capture the difference in different dike-pond types, and was more suitable than the sample point test to assess the accuracy of high-resolution images classification. The extraction result showed that there were 14.06 km2 dike-ponds in Longjiang Town. Among them regular ponds accounted for 58.46%, mainly distributed in the central and northern parts of the town. Their pattern exhibited compactness, which was convenient for expanding aquaculture and then increasing income. Irregular ponds accounted for 41.54%, mainly distributed in the eastern and western parts of the town. Their pattern was less compact, which was conducive to planting and promoted water-land interaction.

Key words: dike-pond, object-oriented classification, Gaofen-2, spatial pattern

中图分类号: