Uses for Photogrammetry on Archaeological Excavations
Photogrammetry is a computerized process that produces spatially accurate images from ordinary photographs. With these georectified images, archaeologists can produce photographic plans of sites and their stratigraphy, take accurate measurements directly from the photo, and import photographic data into other computerized technologies for mapping and visualizing archaeological features. The production of a photogrammetric image involves the combination of a number of technologies: total station surveying, traditional archaeological photography, and geospatial rectification. By combining these technologies, we are able to produce a hybridized documentation technique that can serve many purposes.
What are the benefits?
Traditional photography, while essential for archaeological documentation, has a number of drawbacks. For one, a camera’s lens introduces significant distortion into its images, especially if a wide angle lens is used, therefore no reliable measurements can be taken from a standard photograph. Second, standard photographs do not contain any spatial coordinates, so they cannot be cross-referenced with any geographical data nor included in a Geographic Information System (GIS). Finally, a photograph without any additional meta-data is simply a 2D representation of a plane, its applications are limited. Photogrammetry eliminates these drawbacks and limitations and opens up new opportunities for how photography can benefit archaeological documentation and research by allowing for the creation of spatially accurate two-dimensional images – in both the horizontal (plan-view) and vertical (section-view) planes.
The importance of these spatially precise images is evident when considering their many applicaons. Comparing the original and georectified photos side by side helps illustrate how much effect the processing actually has on the photo. While there may not appear to be anything fundamentally wrong with the first image, the georectified image has a significant decrease in lens-distortion and the planar angle has been corrected. It would not be possible to conduct any accurate spatial analysis on the first image, but many types of analysis can be conducted on the second image.
Photogrammetric images not only provide the viewer with a real birds-eye view of an excavation, they give archaeologists a new way to efficiently document an excavation. Manual drafting is still an essential part of the archaeological process, but photogrammetry adds an additional layer of precision and detail to the archaeological record. Among a number of other uses, a properly georectified image can be digitally traced to produce a plan similar to one drawn by hand.
The georectified images can be layered with these drawings to create a hybrid map of an archaeological site, in much the same way that online mapping services, such as Google Maps, layer street maps with georectified satellite imagery. Additionally, photogrammetry can be used to document sections and any other two-dimensional plane in a spatially accurate photograph. Of course, once photogrammetry imbues an image with spatial correction and coordinates, these images may be imported into a wide variety of software for numerous applications.
The JVRP has incorporated photogrammetry into the recording of all architecture at the square level and the photographic documentation of all sections. An explanation of the workflow that the author (Prins) developed for the JVRP to document architecture will help illustrate some of the applications of photogrammetry.
The photogrammetric recording of an architectural unit in a square is a straight forward procedure. First, a series of markers, or control points, are placed on important parts of the excavation unit, generally this is the architectural feature that is the subject of the photography and mapping. The markers should be brightly colored so they contrast enough with the subject of the photo to be easily differentiated, but small enough to not have a negative visual impact on the image if it were published at a commonly used scale (1:50, for example).
Fewer are necessary in “empty” spaces being photographed, though these spaces should still have collinear coverage. In open spaces within the square which consist only of earth or non-solid elements (like floors), the tabs can be put on the heads of nails pushed into the ground. The author has found surveyor’s Mag Spikes to be useful for this purpose.
2.Total Station Recording of Control Points
After the markers are placed, points are taken directly on them with a total station. This assigns spatial information to the markers in the form of x, y, z coordinates which can be directly viewed in programs like AutoCAD and ArcGIS.
After the points are recorded, a photograph of the area is taken from above, with the lens of the camera roughly parallel to the subject. For this step, the JVRP employs a modified painters’ pole with a heavy-duty camera mount (we use the Pole Pixie Professional Adapter + Tilt Mount) and a handheld radio remote control.
This allows the photographer to position the camera up to 3 meters above the square and take the pictures immediately with the remote. A powerful digital camera is important for clear results. For basic photogrammetry any point and shoot camera will suffice, although photographs with a resolution of at least 5184×3456 pixels will produce better georectified results.
Back in the lab, the total station points are manually referenced to the visible points in the photograph using ArcGIS. The points from the total station are exported into a GIS-ready shapefile that can be viewed in ArcGIS and displayed as a layer in the program. Then the photograph is imported into ArcGIS.
Using the Georeferencing tool, each total station point is matched to the corresponding tab in the photograph. This takes roughly 20 to 30 minutes depending on the number of control points recorded. When enough points have been matched, the technician uses ArcGIS to carry out a polynomial transformation on the image, which projects the photograph onto the spatial information gathered from the total station points, and warps the image to correspond to these points.
The result of this transformation is a photograph with minimal lens-distortion placed within a real-world spatial context. This spatial context can be a site’s local grid, latitudinal and longitudinal coordinates, or in the case of the JVRP, the Israeli Transverse Mercator coordinate system. Now, this image can be dropped into any spatially aware software program and be placed automatically in spatial relationship with any other georecfied images, maps, or plans.
Digitizing Plans from a georectified Photo
One of the most beneficial applications of photogrammetry is the ability to digitize a feature directly from a georectified photo. This can potentially eliminate the need for manual hand-drawing, or at least reduce significantly the time involved in planning. The JVRP uses both traditional hand-drawn plans and digitization directly from georectified photos depending on expediencies in the field.
The procedure for digitizing from a georectified photo is very simple and can be done in any vector drawing program like Adobe Illustrator (though this program can create scaled drawings, it doesn’t translate the spatial information) or CAD/GIS program like AutoCAD or ArcGIS (both of which encode spatial data in every point, line, and polygon drawn). The procedure varies slightly according to the program, but basically a “pen”-tool is used to trace directly over the image. The result is a digital drawing with accurate spatial information.
There are two distinct advantages of digitizing directly from the georectified photo. First, the hand drawing step is eliminated – even a hand drawing will eventually have to be digitized for publication. Eliminating this step reduces errors that occur when a drawing is made from a drawing. Second, the drawing from the georectified image is much more precise, avoiding the potential for overestimation and error inherent in taking hand measurements and translating those to drawing paper.
Site-wide Plans and Stratigraphical and Architectural Photo-mosaics
The procedures described above, produce both georectified spatially correct photos and plans of individual architectural elements. Since these products are encoded with spatial information in the form of coordinates in a known geographic coordinate system, they can easily be ploed together in programs such as AutoCAD and ArcGIS.
This means with simple importing into these programs, one can create site-wide plans combining any of the photogrammetric products. Additionally, photomosaic plans may be compiled from the orthophotos of each architectural element. Photogrammetry, therefore, is not just a means to an end (i.e. the speedy creation of traditional digitized architectural plans), but is also a valuable photographic record of the excavated remains and a powerful tool for visualizing the stratigraphic record.
Photogrammetry of Sections
Along with architectural features, documenting the vertical stratigraphy of archaeological sites is an essential component of archaeological recording. Just as photogrammetry documents horizontal surfaces, it can also be used to document vertical sections with undistorted spatially accurate photographs and section drawings digitized directly from these photographs.
The process works under the same principles as the plan-view photogrammetry described above. Traditionally, documentation of a fully exposed section is typically accomplished by a hand drawing that is later digitized (as with the traditional architectural planning).
However, a georectified photograph can be taken of the section, and a digitized drawing made from it, providing both a spatially accurate photographic record and a digitally-drawn record. In order to produce a georectified photograph of the vertical plane, markers similar to those mentioned above, are placed along the section.
After the markers are placed, their spatial information is collected with a total station. This is a good opportunity to use the total station’s reflectorless mode, which allows the points to be collected without using the prism. This mode is particularly useful when collecting points along a vertical section, since positioning the prism is often problematic. After the points are collected, the section is photographed from ground level, as directly as possible.
Because Georectification software “thinks” in a topographical coordinate system, the x, y, z, values of the points on a section must be dealt with separately. If they are not, the Georectification software will attempt to “look” at the image from the top – i.e. as though it were looking at a physical photograph from above with the photograph standing on its edge. The x, y, z, points, must be converted so that the Georectification software “thinks” that looking directly onto the vertical photograph is the same as looking down at a horizontal plane – i.e. the z-value (up and down) should become the y-value (north and south) in the converted coordinate system.
This could, in theory, be done manually, converging and typing in the coordinates for each in the total station (or in the total station’s exported files), but this would be me consuming. One simply imports the total station output file into this application and the utility outputs a new total station file with the converted coordinates. The new coordinate file and the photograph of the section can now be imported into ArcGIS and the points can be matched up to the visible markers in the photograph according to the technique of a horizontal photogrammetry job.
This produces a georectified photograph .This georectified section photograph can then be used on its own as an accurate representation of the section, it can be used to add details to the hand-drawing, or if a drawing was not originally made in the field, the photograph can be traced to produce an original drawing. This tracing can be done digitally or by hand, using the same method as digitizing architecture from a georectified photograph. The digitized drawing can then be used on its own, or it can be visualized as a superimposed layer on the actual photograph by applying transparency to the digitized vector lines.
Bitelli, G., et al. 2004. Low-Height Aerial Imagery and Digital Photogrammetrical Processing for Archaeological Mapping. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences 34, 55-9.
Bitelli, G., et al. 2004. Low-Height Aerial Photogrammetry for Archaeological Orthoimaging Production. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences 34, 55-9.
Farjas, M. 2009. Digital Photogrammetry: 3D Representation of Archaeological Sites. Ingeniería Cartográfica, Geodésica y Fotogrametría [Online]. Available: hp://ocw.upm.es/ingenieria-cartografica-geodesica-yfotogrametria/3d-scanning-and-modeling/Contenidos/Lectura_obligatoria/photogrammetry1.pdf.
Matsumoto, K. & Ono, I. 2009, Improvements of archaeological excavation efficiency using 3D photography and Total Stations. In: 22nd CIPA Symposium, October 11-15, 2009, Kyoto, Japan.
Osrin, Peter. 2012. TS_Seconer. Megiddo Soware Applications.
Skarlatos, D. & Rova, M. 2010, Photogrammetric Approaches for the Archaeological Mapping of the Mazotos Shipwreck. In: 7th International Conference on Science and Technology In Archaeology and Conservation, 7-12 December, 2010, Petra.