Using photogrammetry on digital x-ray images

Photogrammetry is a technique where digital photographs are converted into 3D models. This method is becoming increasingly popular among archaeologists at NTNU Vitenskapsmuseet, as a way of documenting structures and whole excavation sites in three-dimensional models. To make the 3D model, the object or site to be documented is photographed from several different angles, using a normal digital camera. The photos are then processed in a software program and transformed into a 3D model. The models in this blogpost where made using Aegis Photoscan. The 3D model can among other things be rotated at will, and – if georeferenced – measures can be taken. The model can be converted to PDF format, and thus easily publicized or shared by e-mail.

3D model of an excavation site of an iron slagg deposit in Holm, Rauma, Norway. Photos and 3D model by Dag-Øyvind H. Engtrø.

Screenshot of 3D model of an excavation site of an iron slagg deposit in Holm, Rauma, Norway. Photos and 3D model by Dag-Øyvind H. Engtrø.

The increased awareness of the possibilities of photogrammetry led to the idea of trying this technique on digital x-ray images; could it be possible to make a 3D x-ray image? The tests were performed by the author, conservator Marte Iversen Rønning, in collaboration with archaeologist Dag-Øyvind Hanssen Engtrø. H. Engtrø performed the software imaging and the x-ray images were taken by I. Rønning.

When a digital camera is used to take pictures for photogrammetry, the object/site remains stationary and the position of the camera is changed. The x-ray equipment used during the tests is stationary, so it has to be the object that changes position. The object, in this case a Viking age oval brooch, was rotated slightly for every exposure. All in all there were about 20 x-ray images taken of the test object. The same exposure values were used for all the images of one object.

The 3D model of the oval brooch shows the shape of the object, and the positioning of the decoration quite well, without the distortions that can happen in regular 2D x-ray images. The model is made from 19 images taken from different angles, but all of them are taken of the brooch’s upper side. It was not photographed from below due to time restraints and the fact that it was only done to test the technique.

To see the 3D model follow the link below. It might take a little while to open. It starts off with the brooch facing the wrong way, so it looks blank, but if you click your mouse over it and drag, the 3D image will appear, and you can move it around as you please.

https://www.ntnu.no/documents/10476/1263589258/3D-modell+r%C3%B8nthen+st%C3%A5lspenne/debfd362-8931-4edd-91c3-1ba8e3d8647d

Attempts were made to make 3D images of some other archaeological objects, but the results have not been very successful. This is probably because the rotation of the objects was not optimal to produce a good 3D image. This has to be tested further.

The most useful application of photogrammetry on x-ray images could be for block lifts from archaeological excavations. Block lifting is a technique used to remove fragile or complex objects from a site in a block of soil to continue excavating them in the conservation lab. Using photogrammetry may allow the conservator a better understanding of the spatial distribution of objects within the block lift.

One of the noticeable drawbacks of using photogrammetry on x-ray images is the amount of time it takes to produce the x-ray images for one object. To get a good 3D image, you will probably need at least 20 images, but for important or intricate objects this might be worth the time. The technique may not be suitable for unstable objects since it is required that objects are rotated.

The software also requires some training to use, and the cost of it is quite high, but this will likely change with time. The software has already become much more user-friendly over the last couple of years.

There seems to be great potential for using photogrammetry on digital x-rays, and also in conservation generally. We would love to know if anyone else has tried using this technique. We also encourage you to try it for yourselves.

 

The different views of the object used to make the 3D model.

The different views of the object used to make the 3D model.

Be first to comment