Introduction of HTGCL (Historical Topographic Ground Control Line) - New paradigm of handling historical map in GIS - /FOSS4G Japan 2020

Introduction of HTGCL (Historical Topographic Ground Control Line) - New paradigm of handling historical map in GIS - /FOSS4G Japan 2020

Slide for FOSS4G Japan 2020 presentation.

E83ff81bc5a881c33dcff37eb160ce12?s=128

Code for History

November 08, 2020
Tweet

Transcript

  1. Introduction of HTGCL (Historical Topographic Ground Control Line) - New

    paradigm of handling historical map in GIS - Kohei Otsuka – Code for History 1
  2. Table of contents • What is Code for History? •

    Introduction to • Introduction of HTGCL (Historical Topographic Ground Control Line) and its application to GIS • Other small topic (Not related with HTGCL) • Conclusions 2
  3. What is Code for History? 3

  4. A community that uses information technology to solve historical problems

    (practically one person) https://code4history.dev/ 4
  5. Characteristics • I want to emphasize that for us, information

    technology is a "means" and not an “objective". • The "objective" is to solve a historical problem, and if necessary, to utilize non-IT means. • The first priority is always to solve the problem. 5
  6. Introduction to 6

  7. Map viewer technology that allows you to enjoy walking around

    the city using historical maps and pictorial maps © OpenStreetMap © Utouto Tanken-tai 7
  8. We recently changed it’s logo 8 Maplat

  9. Features of • Not distorting historical maps • Adjusting not

    only the location of the center point, but also the direction and scale • Bidirectional one-to-one conversion (Returning to the original display position after multiple map changes) • Converting lines to lines (Roads to roads, rivers to rivers, etc.) • It's open source, so anyone can create their own site with it • Project Sites: Library: https://github.com/code4history/Maplat/wiki Editor: https://github.com/code4history/MaplatEditor 9
  10. Not distorting historical maps • Existing technology (GIS) distorts historical

    maps 10
  11. Not distorting historical maps • overlays historical maps without distorting

    them 11
  12. Adjusting the direction and scale • Existing technology (Stroly) can't

    adjust the right orientation and scale 12
  13. Adjusting the direction and scale • fits the direction and

    scale almost perfectly 13
  14. Bidirectional one-to-one conversion (homeomorphic conversion) • Existing technology (Stroly) shifts

    the display position when continuously switching maps 14
  15. Bidirectional one-to-one conversion (homeomorphic conversion) • does not shift the

    display position when continuously switching maps (Japan patent JP-6684776) 15
  16. • In the past, used to get misaligned on the

    road at times. Converting lines to lines Modern map Historical map 16
  17. • In the past, used to get misaligned on the

    road at times. Converting lines to lines Modern map Historical map 17
  18. • Latest can accurately convert a point on the road

    to a point on the road Converting lines to lines Modern map Corresponding lines (HTGCL) Historical map 18
  19. • Latest can accurately convert a point on the road

    to a point on the road Converting lines to lines Modern map 対応線 Historical map 19 Corresponding lines (HTGCL)
  20. Introduction of HTGCL (Historical Topographic Ground Control Line) and its

    application to GIS 20
  21. What is the HTGCL? • Abbreviation for Historical Topographic Ground

    Control Line • Naming of the corresponding lines between modern geographic coordinates and historical map coordinates which are used in 's "converting lines to lines" function. • The concept is borrowed from Ground Control Points (GCPs), which are the points in existing GIS that correspond to inaccurate maps and accurate maps. • Relying on points alone for coordinate transformations between maps makes it difficult to maintain homomorphism throughout the transformations, and it is not possible to convert lines to lines. Therefore, transformations using HTGCL are valuable not only for but also for existing GIS. 21
  22. Analyzing and GIS processes • 's function can be divided

    into two main processes • Converting coordinates between exact geographic coordinates and historical map coordinates, with features such as maintaining homomorphism and converting lines to lines => A • Overlapping historical map and modern map according to the result of the coordinate transformation, with adjusting the position, direction and scale. => B • Handling historical map with existing GIS also can be divided into two main processes • Converting coordinates between exact geographic coordinates and historical map coordinates => C • According to the results of the coordinate transformation, the historical map is distorted, reorganized into precise geographic coordinates and superimposed on the modern map. => D • If 's coordinate transformation A is better than existing GIS's coordinate transformation C, can't 's coordinate transformation A can be exported as pre-process of D in GIS? 22
  23. Comparison of HTGCL and existing GCP in coordinate conversion Method

    Rely on Pros Cons GIS (GDAL, Thin plate spline) GCP • Relatively easy to create data • Smooth conversion results • Conversion is not continuous or bidirectional. • Not converting lines to lines HTGCL • Homeomorphic conversion • Converting lines to lines • Conversion becomes sharp on the edges of the triangle network • Resolving topological errors manually is a pain 23
  24. An example of application of HTGCL transformation to an existing

    GIS 24 The result used GCPs (GDAL) The result used HTGCL
  25. Homeomorphic transformation 25 The result used GCPs (GDAL) The result

    used HTGCL • The continuity and bi- directionality of the map coordinates is lost, as evidenced by the occurrence of spatial loops in some places. • Although intensely distorted, the continuity and bi- directionality of space, or in other words, homeomorphi sm, is strongly maintained.
  26. Transforming lines to lines 26 The result used GCPs (GDAL)

    The result used HTGCL • The green line shows the modern roads, but the roads on the historical map side have been violently distorted into a curve and are off the green lines. • The roads on the historical map side have been transformed to follow the green lines definitely, although somewhat distorted.
  27. Cons: Sharp transformation on the edge of triangulation 27 The

    result used GCP Original data used in conversion • Transformations using HTGCL are computed by vector transformations within triangles generated by the triangulation. • As a result, the transformations across the edges can result in sharp shape. • We are investigating whether there are any measures to mitigate this change. Sharp conversion The edge of triangulation which is used for this sharp transformation
  28. Cons: The pain of manually fixing topological errors 28 Topological

    errors case Fixed by adding dummy GCP Errors Dummy GCP • To perform a homeomorphic transformation with HTGCL, there must not be any topological errors in the triangulation. • If topological errors exist, it can be resolved by adding dummy GCPs in the appropriate places, but for now we have to do it manually. • We are collecting enough examples of what error cases and where GCPs need to be added, and hope to automate topological errors resolution in the future.
  29. HTGCL based map transformation function will be in MaplatEditor 29

    • Currently, there is no solution for converting historical maps using HTGCL to GIS data such as WMTS. • However, in the near future, MaplatEditor will include the function to generate WMTS tile data from HTGCL. • In addition to HTGCL support, we are also considering various functions to handle WMTS, including support for Japanese old GSI topographic map tiling and the creation of new tile map that slightly shifts the coordinate system of existing tile maps. • In other words, MaplatEditor aims to be a GIS editor product that can be an alternative to QGIS's Georeferencer and GDAL's gdal2tiles.py, rather than just a data editor for .
  30. Other small topic (Not related with HTGCL) 30

  31. PWA service worker framework for caching tile map 31 •

    Now we are developing a Service Worker framework that works with PWA (Progressive Web App) that can cache tile data for tile maps offline. This framework makes it easy to create PWA map applications that can work offline. • It supports caching of map tiles, bulk download of caches, and clearing of caches. The Service Worker is already almost complete and we are developing a wrap class to make it easy to use on the front end. At the moment it only supports raster tiles, but if it is technically possible to support vector tiles, we will support it. • https://github.com/code4history/TileCacheServiceWorker
  32. Conclusions 32

  33. • is an open-source map viewer technology that allows you

    to walk around town using historical maps. • One of the features of is the ability to convert lines to lines in the coordinate conversion between historical maps and modern maps. • The corresponding lines networks those are set up when converting lines to lines is named HTGCL. • Since coordinate conversion using HTGCL has advantages over coordinate conversion using GCP in existing GIS, there is also value in creating tile map data by GIS-like historical map image transformation. • MaplatEditor will not only serve as a data creation editor for , but will also enhance the ability to create GIS tile map data in the future. Conslusions 33
  34. • I have developed through my experience as a GIS

    engineer, and as the antithesis to historical map processing in GIS. As a result of pursuing that approach, I am very happy about the possibility of providing a new paradigm on the GIS side as a result. • I think this is also the result of developing , a solution with a different concept than GIS, but with the base technology system of GIS in place. I was also previously the developer of another solution called Stroly, but Stroly was a technology that was not based on a GIS technology system, and if I had continued to develop Stroly, I don't think we would be making this presentation today. • However, at any rate, is being developed by just one person, so I have no timetable yet for when this feature will be implemented. I have a lot of other features that I need to prioritize and develop, so it's likely to be years in the future. • is an open source activity, so if any GIS expert developers are interested in this presentation, please help me. Last comments 34
  35. End User: kochizufan Github pages: https://code4history.dev/ E-mail: kochizufan@code4history.dev Slide: https://speakerdeck.com/kochizufan/foss4g-japan-2020

    35