Towards Standards for the Manual Digitization and Vectorization of Analog Historical Maps and Records

While the development of automatic object recognition and vectorization algorithms to be used on scanned or digitized maps is a vibrant area of study today, most are not developed for or suitable for use with either historical maps in general or archaeological field maps both historic and modern. Current algorithms use methods such as line thinning and other approximation techniques while also relying on continuous lines and features as well as color/greyscale spectrums in the source maps to function. These often are not present in hand-drawn field maps or very old maps which can contain many irregular features and symbols and often have no exploitable color either due to fading or damage associated with age or simply by not being drawn with any. Current automatic algorithms also require manual intervention and error correction even in cases in which they do work. This has driven many projects involving historical maps to use manual vectorization methods instead however in very few cases is there any mention to the specifics of this process. This lack of a standard method for manual vectorization makes the results of such projects hard to compare. This study proposes one possible set of standards for manual vectorization of historic maps and archaeological field maps of all eras with the hopes of further refinement and improvement to be used by future studies. Using the field maps from the University of Iowa Archaeological Field School site at Woodpecker Cave (13JH202) rock shelter in Coralville, Iowa from the 2012-2015 seasons, the proposed methodology is explained and demonstrated. After scanning the field maps and georeferencing them to a vector representation of the excavation site and its units, the numbered artifacts in each unit are traced and converted into vector polygons. A quantitative comparison is then made between the new polygons, their centroids (center points), and theodolite point locations of the artifacts taken in the field to determine how far off the drawn artifacts are from their “true” locations of the theodolite points. A brief statistical analysis of the results follows along with a discussion of the value of this process. It is shown that the artifact polygons not only are a valuable supplement to the site’s field records but that the vector format enables the artifacts of the site to be analyzed in a more complex way using GIS (Geographic Information System) software.