Digitize historical maps with raster-to-vector conversion and use them in GIS

If you digitize old paper maps (historical maps) and import them into a GIS (geographic information system), you can freely compare and analyze past and present geographic information. Historical maps come in various types such as cadastral maps, old edition topographic maps, and Edo-period pictorial maps, but as analog materials they are hard to use and are at risk of deterioration. By digitizing historical maps using raster-to-vector conversion, you can efficiently convert lines and shapes drawn on paper into vector data and overlay them with modern map data for use. This article explains in detail the process of digitizing historical maps, the role of raster-to-vector conversion, and how to utilize the resulting data in a GIS.

Digitizing historical maps brings significant benefits across various fields. For example, it is useful in the following situations:

• Geographers and researchers: You can analyze changes in past topography and land use. By overlaying historical maps with current maps, you can closely track historical geographic changes such as river course changes and urban expansion.

• Municipal disaster prevention officers: If you digitize historical flood inundation extents or disaster history maps and overlay them on current maps, you can identify areas with high disaster risk. Comparing past disaster records with current maps can improve disaster prevention planning and the accuracy of hazard maps.

• Surveying and design engineers: If cadastral maps or old design drawings that remain on paper are converted into vector data, they can be handled directly in CAD or GIS. You can integrate existing boundary lines and control point positions into the local coordinate system and use them as reference materials for new designs.

Now, let’s look at the specific procedures for digitizing historical maps and the key points of raster-to-vector conversion.

What is raster-to-vector conversion

First, what is raster-to-vector conversion? Raster-to-vector conversion is a technique that automatically converts a map captured as raster data (image data) into vector data (line and point data). When you scan a paper map, it becomes an image (raster), but as it is, lines and text are just collections of pixels. Using raster-to-vector conversion software, the software analyzes features such as roads and boundary lines in the image and connects continuous pixels to extract digital lines (vectors). In other words, it automatically traces the lines on the image and creates drawing data usable in CAD or GIS.

The advantage of raster-to-vector conversion is that it is far more efficient than manually tracing from scratch. Digitizing large-area historical maps or complex drawings by hand takes enormous time and effort, but using dedicated software you can obtain vector data with a certain level of accuracy in a short time. Raster-to-vector conversion software is particularly useful for materials such as cadastral maps and old topographic maps that have clearly printed lines.

However, there are caveats to raster-to-vector conversion. The accuracy and quality of automatically converted data depend on the source image. If the scanned image is unclear or if folds and distortions cause non-uniform scale, the positional accuracy of the conversion results will be poor. Also, text and symbols differ from lines, so they may not be recognized correctly and can become distorted. Although raster-to-vector conversion software performance improves yearly, manual correction and verification after conversion are still indispensable. For example, you will likely need to補完細部の線 that could not be fully converted or re-enter place names and numbers that were crushed.

Even considering these limitations, raster-to-vector conversion plays an important role in the process of digitizing historical maps. Especially when handling large numbers of paper maps, combining appropriate preprocessing with raster-to-vector conversion makes it possible to efficiently produce vector data. Next, let’s look at the specific steps for digitizing historical maps.

Steps to digitize historical maps

To digitize historical maps, several steps must be followed in sequence. Here we introduce the general workflow.

1. Scanning the historical map (rasterizing)

First, scan the paper map with a scanner or a high-resolution camera to create digital images (raster data). It is important to scan at as high a resolution as possible so that fine lines and text on the map are legible (generally 300–600 dpi is recommended). For large-format maps, use a large-format scanner or scan multiple sections and stitch the images together later. Also, before scanning, it is important to flatten folds and creases as much as possible. Old maps may be distorted from being folded, so pressing them flat with a transparent weight (such as a glass plate) while scanning can yield better results. Use non-lossy or lossless formats like TIFF or PNG for scanned images to preserve clarity without degradation.

2. Georeferencing the scanned image

The scanned map image is not yet tied to any coordinate system. In other words, to compare or overlay it with modern maps, georeferencing is necessary. Georeferencing is the process of assigning geographic coordinates such as latitude/longitude or planar coordinates to specific points on the image and fitting the entire image to actual geographic space. Specifically, select multiple known points on the historical map (for example, intersections, building corners, or triangulation points still identifiable today), obtain their coordinates on the current map, and link them. By entering a number of control points into GIS software, the software will apply appropriate rotation, translation, and scaling (and, if necessary, nonlinear distortion correction) to align the historical map image with a modern coordinate system.

This alignment process ties the scanned historical map to real geographic space so it can be correctly overlaid with other GIS data (modern maps, aerial photos, etc.). To improve georeferencing accuracy, it is desirable to distribute reference points evenly across the map extent. If the historical map contains a latitude/longitude grid or coordinate markings, use them to achieve better alignment. For pictorial maps from before the Meiji period, precise alignment may be difficult, but roughly overlaying major landmarks can still be useful for comparative purposes.

3. Preprocessing the raster data (image adjustment)

Once georeferencing is done, perform preprocessing to improve the scanned image before raster-to-vector conversion. First, for color or grayscale historical maps, consider converting them to black-and-white binary images. Raster-to-vector conversion software usually works best when lines (black) and background (white) are clearly separated. Photographic-like shading or paper discoloration and stains can be interpreted as noise if left as-is. Use image-editing software to adjust contrast and remove unnecessary colors so that only the map linework stands out.

If the scanned image contains extra frame lines, margins, or skew from scanning, perform trimming and rotation correction at this stage. Consider using filters for noise reduction as needed. However, applying overly aggressive filters may remove fine roads or boundary lines, so adjust carefully. The goal of preprocessing is to provide as crisp and low-noise a binarized image as possible so the raster-to-vector algorithm can accurately detect map lines.

If lines are thick and blurred, perform a thinning process (image processing that narrows lines) so the conversion can more easily capture the line center and reduce the occurrence of double lines.

4. Performing raster-to-vector conversion (automatic vectorization)

When preparations are complete, load the image into raster-to-vector conversion software and run automatic vectorization. The exact workflow depends on the tool, but typically you open the image file, set conversion parameters, and start processing. Parameters include thresholds for what to recognize as a line (how dark a pixel must be to be considered black), size thresholds for areas to be treated as noise, and the precision for approximating curves with straight or curved segments. Appropriate parameter settings will yield better results.

When the conversion starts, the software analyzes lines in the image and generates vector objects (points, lines, polygons, etc.). Roads and boundaries, for example, will be output as continuous polyline data, and building outlines as polygons. Choose output formats compatible with CAD/GIS such as DXF or SHP according to your purpose. Processing time depends on image size and content, but compared to manual work it is much faster; even large maps can be processed in minutes to tens of minutes.

Review the automatic conversion results and remove obviously unnecessary vector elements (garbage or misrecognized lines). If building names or place names have been extracted as unnatural lines, delete those line segments and, if needed, record text information separately as text data. Some raster-to-vector tools allow you to exclude specific colors or areas during conversion, so masking text areas before running the conversion can be an effective strategy.

5. Editing, integrating, and saving vector data

Edit and clean up the vector data obtained from raster-to-vector conversion as needed. When joining multiple maps, check for misalignment at seams and make fine adjustments if necessary. Even data exported from a single historical map may have broken lines or extra vertices. Use GIS or CAD software to reconnect lines that should be continuous and to remove duplicated segments. It is especially important that boundary lines and road networks are continuous.

Once editing is complete, save the data in an appropriate format. For GIS use, exporting to Shapefile (SHP), GeoJSON, or a database format makes it easier to integrate with other geographic data. For CAD use, saving as DXF lets you combine the vectors with other drawings or repurpose them in design. When saving, be sure to confirm the coordinate system and unit settings to prevent misalignment when overlaying with other data.

By following these steps, paper historical maps are reborn as digital vector data. Next, let’s look at concrete examples of how such historical map data can be used in a GIS.

Using historical map data in GIS

Vectorized historical maps obtained via raster-to-vector conversion can be used in various ways in modern GIS environments. Here are representative use cases for historical map data.

Use in geographic and historical research: comparative analysis of past and present

Digitized historical maps are useful for comparative analysis of past and present geography when overlaid with modern map data. For geographers and historians, insights gained by comparing historical maps with current maps in a GIS are numerous. For example, overlaying river courses or lake shapes from Meiji-era topographic maps with current satellite imagery can clearly show that areas where rivers once flowed are now occupied by residential developments. Similarly, comparing old city maps with modern road networks can reveal how former highways or railway lines that were abolished or relocated influenced current urban structure.

Visualizing historical maps in GIS can reveal geographic patterns not obvious from paper sources. For example, matching village layouts depicted in Edo-period pictorial maps with modern topographic maps may reveal the environmental conditions (distance from rivers, high or low ground, etc.) considered by people when forming settlements. Overlaying maps from multiple periods allows quantitative assessment of long-term environmental changes such as urban growth rates or forest loss/gain. Such comparative analyses are valuable as materials for papers and reports and can present persuasive narratives when shown as visual figures.

Use in disaster prevention and hazard mapping: visualizing risk areas

Historical map data is also valuable for municipal disaster prevention planning. By reflecting past disaster histories and topographic information onto current maps, you can identify potential risk areas. For example, if you extract old river channels depicted in early Showa-era topographic maps as vector data and overlay them on modern residential maps, you can determine whether current housing is built on former riverbeds. Areas developed on reclaimed land or former river courses may have higher risks of liquefaction and inundation and should be treated with caution in disaster planning. Likewise, plotting place names such as “— embankment” or “swamp” from Edo-period maps in GIS can provide clues to terrain that was historically prone to flooding.

Moreover, digitizing old hazard maps or disaster record maps and comparing them with current maps can assist in disaster planning. For example, overlaying historical landslide extents on current cadastral maps to identify which modern districts overlap with past events can highlight priority areas for hazard mitigation. Adjusting the transparency of historical map layers in GIS makes it easy to intuitively perceive topographic features that are no longer visible (former cliff lines, wetlands, etc.). Using historical maps in disaster prevention can thus improve the accuracy of hazard maps and help create persuasive educational materials for residents.

Use in surveying and design: integrating existing assets data

In surveying and civil engineering design, converting old drawings and cadastral maps to vector data via raster-to-vector conversion can be applied to modern planning tasks. For instance, digitizing cadastral maps (public maps) preserved on paper and importing them into GIS allows overlay with current cadastral survey data and aerial photos to examine changes in land boundaries. Comparing boundary lines derived from old cadastral maps with current registration information can indicate whether discrepancies exist, which may prompt re-surveying on site.

In design work, past drawings are often referenced. Vectorizing old plan drawings or facility layout maps with raster-to-vector conversion lets you use the positions and shapes of existing structures as a base for new designs. Managing old piping diagrams or road drawings in a GIS can reveal mismatches during renovation planning and help prevent design errors. If existing drawings are digitized, it is easy to extract only the necessary parts for reuse in new drawings. This is particularly useful in remodeling or expansions, where directly reusing existing facility geometry improves work efficiency.

Combining digitized historical map data with survey results on site (GPS survey points or electronic reference station coordinates) enables smooth integration of past survey maps with current coordinate systems. This allows you to reproduce benchmark or leveling point information depicted in old drawings on modern base maps and reflect them in design and construction.

Accuracy verification and correction using on-site surveying with LRTK

When using historical map data in GIS, it is ideal, if possible, to cross-check and adjust accuracy with on-site survey measurements. A useful tool for this is the increasingly popular simple surveying system LRTK. LRTK is a system that links a compact high-precision GNSS (satellite positioning) receiver with a smartphone or similar device to enable real-time centimeter-level positioning (cm level accuracy (half-inch accuracy)). Compared with traditional surveying instruments, LRTK is highly portable and offers intuitive operation for acquiring position information, making it easy to use even for non-specialist surveyors.

With LRTK, you can easily obtain precise current coordinates for points depicted on historical maps. For example, measure the positions of benchmark points or important intersections shown on historical maps using LRTK on site, then compare those coordinates with the corresponding points on the already-georeferenced historical map data to verify the positional accuracy of the digitized data. If discrepancies are found, you can use the measurements to make fine adjustments in GIS—such as translating the whole dataset by several meters (several ft) or applying local rubber-sheet corrections—to improve consistency between the historical map data and the on-site coordinate system.

Because LRTK surveying is fast, you can collect additional data in the field as needed. Measure current conditions not present on historical maps (positions of new buildings or topographic changes) and import them into the same GIS project to create comprehensive map materials that integrate past and present data. By combining raster-to-vector-converted historical map data with on-site LRTK survey data, the reliability and practicality of the digital maps are significantly enhanced. Going forward, incorporating affordable and easy-to-use LRTK in various projects will likely advance the precise utilization of historical materials.

Conclusion

This article explained the significance, procedures, and applications of digitizing historical maps using raster-to-vector conversion and utilizing them in GIS. By scanning paper maps, georeferencing and preprocessing them, and performing raster-to-vector conversion, vast amounts of historical information can be revived in modern digital mapping environments. The resulting vector data can be applied across a wide range of fields—from geographic research to disaster prevention to surveying and design—becoming valuable datasets that connect past and present.

Although digitizing historical maps may initially seem laborious, it can be done efficiently by following appropriate procedures. In particular, using the automation provided by raster-to-vector conversion greatly reduces processing time and enables the digitization of more materials. Combining new surveying technologies such as LRTK further increases the accuracy and value of the digitized historical map data.

Historical maps contain many insights applicable to the present. By digitizing them and analyzing them in GIS, we can learn from the past and apply that knowledge to contemporary society. Use raster-to-vector conversion to explore the world of historical maps in digital space, and leverage accumulated regional records and wisdom for future disaster prevention, regional planning, and academic research.