历史地理研究 ›› 2022, Vol. 42 ›› Issue (2): 117-133.

• 专题研究 • 上一篇    下一篇

基于机器学习和图像形态学的彩色近代地图数字化——以近代上海地区地表水体信息提取为例

柴宝惠   

  1. 复旦大学历史地理研究中心,上海 200433
  • 收稿日期:2021-09-18 出版日期:2022-05-20 发布日期:2022-08-31
  • 作者简介:柴宝惠,女,1992年生,辽宁沈阳人,博士,复旦大学历史地理研究中心博士后。
  • 基金资助:
    国家社会科学重大项目“魏晋隋唐交通与文学图考(18ZDA247);中国博士后科学基金会面上资助项目“近百年来上海城市化进程中的地表水体变迁研究(1920—2020)”(2021M690671)

Digitization of Old Maps Based on Machine Learning and Image Morphology: an Example of Surface Water Extraction in Modern Shanghai

Chai Baohui   

  1. Center for Historical Geographical Studies, Fudan University, Shanghai 200433, China
  • Received:2021-09-18 Online:2022-05-20 Published:2022-08-31

摘要:

利用科学测绘技术绘制的近代地图作为一类珍贵的历史地理资料,不同程度上反映着过去的地表覆盖情况,数字化则是复原地图所载地表覆盖信息的重要途径。以《华东·上海》地图为例,实现并验证一种基于机器学习和图像形态学的彩色近代地图数字化方法。结果表明,该方法能够充分挖掘地图中的颜色信息和形态结构信息,以半自动方式快速准确地将彩色近代地图中的地表水体信息提取出来。该方法对数字化一类彩色近代地图具有参考价值,有望为精准复原近代以来地表覆盖变迁,深入理解人地关系变化提供数据和方法基础。

关键词: 彩色近代地图数字化, 近代上海, 地表水体, 机器学习, 图像形态学

Abstract:

Modern maps drawn using scientific surveying and mapping technology are precious historical geographical data source, directly showing land cover information in the past. Digitization is a necessary way to extract the historical land cover and its changes from old maps. This paper proposes an old map digitization method based on machine learning and image morphology analysis, and takes the “Eastern China: Shanghai” map as an example to elaborate its implementation process and verify the effectiveness. Results show that the method can fully utilize the color information and morphological information in the map, and extract the surface water information quickly and accurately in a semi-automatic manner. The proposed method can be applied to the digital extraction of land cover information with color characteristics in most color modern maps. It shows great potential to provide both data and method basis for accurately restoring land cover changes and understanding the evolution in man-land relationship since modern times.

Key words: digitization of old maps, modern Shanghai, surface water bodies, machine learning, image morphology

中图分类号: 

  • K928