Historical map polygon and feature extractor
Historical map polygon and feature extractor
复制标题
历史地图多边形和特征提取器
DOI：
10.1145/2534931.2534932
复制
发表时间：
2013
影响因子：
8.2
通讯作者：
M. Arteaga
中科院分区：
文献类型：
--
作者：
M. Arteaga
从19世纪和20世纪初的美国保险地图集等历史地图中提取多边形和属性数据迄今为止一直是一项手工任务。纽约公共图书馆（New York Public Library， NYPL）目前依靠员工和志愿者的工作，从馆藏中手动提取多边形和其他属性数据，为城市历史研究创建新的公共数据集。这是一项耗时的任务，需要长达几分钟的时间来跟踪shapefile并转录单个建筑物的属性。本文提出了一种从历史地图中自动提取此类属性数据的方法。该方法利用多个图像处理和统计实用程序，在手工操作所需的一小部分时间内产生理想的结果。
Polygon and attribute data extraction from historical maps such as US insurance atlases from the 19th and early 20th centuries has so far been a manual task. The New York Public Library (NYPL) currently relies on staff and volunteer work to manually extract polygons and other attribute data from its collection to create new public data sets for the study of urban history. This is a time-intensive task requiring up to several minutes to trace a shapefile and transcribe attributes for a single building. In this paper we propose an approach to automatically extract such attribute data from historical maps. The approach makes use of multiple image processing and statistics utilities to produce desirable results in a fraction of the time required to do by hand.
On average, a shapefile for an atlas sheet is generated in ~11.4 minutes for a total of 23.5 hours of processing time for a whole atlas that contains ~55,000 polygons; contrast this time frame to NYPL's current manual process that has taken three years to extract about 170,000 polygons across four New York City street atlases.
Even with some error rate in the proposed approach, the most cumbersome, time-intensive work (manual polygon drawing) has been reduced to a fraction of its original scope. This new workflow has promising implications for historical GIS.