TRACCE no. 11 – by Elizabeth Donnan
Despite the recent upsurge of interest in rock art studies relatively little research has been conducted into the reasons behind the techniques used to record discoveries in the field. At present the techniques in use have been defined and influenced by current issues of interest which focus on interpretation and meaning with little regard to recording methodology…
Viewing recording as a secondary aspect in this field of study can only be detrimental, especially now, in an age when emphasis is laid heavily on the reliability of stratigraphic and contextual citations. Therefore, recording, as a basis for and essential form of fieldwork, needs to be stressed in a climate where more and more information is amenable to computer storage, analysis and manipulation (1).
The aim of this paper is to discuss the application of digital image processing (DIP) as a new and non-destructive technique for the recording of British rock art. After a brief review of the previous research conducted in British contexts I will subsequently outline the results of fieldwork undertaken in August 1998 to record a known rock art motif in Northumberland, north-east England. My discussion will question the applicability of such a method to recording contexts in Britain and the theoretical problems behind such an approach.
Background of previous research
Northumberland contains the largest number of recorded cup-and-ring marked sites in Britain. Recordings of motifs began in the 19th century with the work of antiquarian researchers Tate (2) and Collingwood Bruce (3)). However, little subsequent work was undertaken to record sites and post-antiquarian research has tended to focus on typology, chronology and meaning, with little concerted effort to publicise recordings of the motifs present (4).
Twentieth century work has produced more systematic studies of rock art sites. This includes work by Morris in Scotland, (5) the Ilkley Archaeological Group in Yorkshire (6), and Beckensall in Northumberland (7). Beckensall has produced an extensive record consisting of rubbings and photographs of all the known marked rocks in the area over the last 10-20 years.
Apart from photography the standard techniques used to record rock art in Britain, therefore, include drawing, tracing and rubbing. These are mainly direct-contact methods which appear to work well. They are thus firmly ingrained in the British recording methodology. Although the destructive nature of these techniques is documented in the international literature on rock art (8) it seems that little has been done to disseminate this information in Britain.
Case study – Weetwood Moor
The sites at Weetwood Moor near Wooler, Northumberland, are evidence of the well marked outcrops available for study. The example discussed reflects the clearly visible peck marks of the motif and the relatively flat surface upon which it appears. It was decided that such a motif would be ideal for an initial testing of DIP techniques. Photography was used to capture the initial image (fig. 1)
This was digitised and processed using the available DIP package at Durham University (PC_IMAGE PLUS) which, unfortunately, only allows for greyscale image viewing (fig. 2). However, this is not seen as a hindrance at this stage of investigative work.
As detailed by Clogg and Díaz-Andreu pre-processing is vital for any DIP work on a rock art motif (9). In contrast to their work with painted motifs, problematic areas with petroglyphs include the recognition of natural and artificial marks on the rock surface, and, more alarmingly, the potential for a total separation of motif from rock surface if such techniques are used without scrutiny.
With this example other problems were highlighted. Firstly, the problem of lichen on the rock surface. Second, the question of lighting when dealing with accurate portrayal. Third, the need to record the peck marks as seen. These are not apparent in previous recordings. Consequently, a variety of processing tools were used in an attempt to extract an outline of the motif despite such problems, and to investigate the potential of further applications in studying the form of the markings.
The use of random colour look-up tables proved helpful in identifying previously unclear marked areas. In fig. 3 the lichen is markedly different in colour and appearance to the rest of the rock surface. Pixel processing over the whole image, including shade correction also proved effective. This involved the subtraction of a processed version of the original from the original image itself. The result emphasises certain features, especially the overall shape and size of the rings for an outline (fig. 4).
Despite reasonable results from whole image processing the parts of the motif obscured by lichen were still difficult to perceive. It was necessary to select certain areas as regions of interest (ROIs) which allowed targeted processing to achieve the clearest viewing of all aspects. Each ROI was thresholded to generate a binary image (pixels over a certain greyscale value become one colour, i.e. pink, and others are changed to 0, i.e. black). With careful pixel manipulation and binary editing (e.g. freehand drawing) a tentative outline could be achieved. This was influenced by the lighting in the original photograph and proved that some form of shading is desirable for further image processing. Fig. 5 shows, however, that binary editing of the thresholded image contains a degree of subjectivity which needs to be carefully monitored. The image shows two separate outline attempts. The first is indicated by the colour blue, the second by the colour pink. The red colour shows the degree of difference and overlap between the two attempts.
Combining the ROI binaries produced the whole binary image (fig. 6). At this stage the rock surface is shown with an overlay outline of the image, however, this can be removed and the binary image viewed in isolation as in fig. 7. This was undesirable as the motif was then viewed entirely out of context. Consequently, I prefer to use the filled outline which, with further editing, can be reduced to a single line representation. In this way the peck marks within the rings are shown (fig. 8). This presentation is thought to be the most representative of the actual motif and further processing will only increase the subjectivity apparent through freehand outlining of the motif.
To study the form of rings and peck marks another function was utilised. This involved line analysis. The result of line analysis was a line profile.Although this charts the greyscale values rather than actual depth it is interesting to consider such an approach in future analysis. This would alleviate the necessity of direct contact with the rock surface. The integrate line function draws several close-set lines to achieve accumulated data. This might improve the accuracy of measurements. The inclusion of a scale allows pixels to be calibrated and metric measurements taken from the line profile. The line profiles for Weetwood are shown in fig. 10a and b.
Although results are patchy at this stage it is possible to state that DIP is a viable option for rock art recording in this country. It is, however, important to consider the theoretical concepts behind such techniques. The possibility of obtaining an outline and line profile, although valuable to accurate recording and research, promotes the potential for typological study. This is alarming since, once again, motifs might be taken out of context and studied as individual elements rather than as aspects of a larger design or ideological element (10). It is important that the shortcomings of DIP are highlighted. These include the inherent level of subjectivity apparent in any attempt to gain an outline of a motif, despite the possibilities of accuracy which are prevalent in computer techniques. It is also worth mentioning that at present DIP depends upon traditional photographs for good quality resolution images. Therefore, photographic techniques must be modified to allow for greater digital manipulation of the image. In my opinion this renders the use of low-angle light and stark shading, as advocated for traditional rock art photography, a hindrance. The photograph must be seen as the starting point for recording rather than an end product.
In Britain DIP is an exciting option for study at a time when new and more integrated approaches are much needed (11). It is hoped that further work with DIP, provided that theoretical concerns are addressed, will allow for a more integrated approach alongside the appropriation of a more complete and accessible database for this country.
Department of Archaeology
University of Durham
- For a more detailed analysis of recording techniques and digital image processing see Donnan 1998.
- Tate 1865.
- in Beckensall 1983.
- Hadingham 1974.
- Morris 1970-71.
- Hedges 1986.
- Beckensall 1983, 1991.
- Mark & Newman 1994.
- Clogg & Díaz-Andreu 1998.
- Mitchell (1992, 7-8) emphasises the post-modern principles in DIP. This includes the possibility of image manipulation leading to a greater isolation of motif from surface than that which already exists, thus allowing for endless interpretations of the same motif or site.
- Bradley (1997, 7) highlights the fact that rock art in Britain is confined to the margins of prehistoric studies.
- BECKENSALL, S. 1983. The Prehistoric Rock Carvings of Northumberland. Pendulum Press.
- BECKENSALL, S. 1991. Prehistoric Rock Motifs of Northumberland vol. 1. Abbey Press.
- BRADLEY, R. 1997. Rock Art and the Prehistory of Atlantic Europe. Routledge.
- CLOGG, P. & DÍAZ-ANDREU, M. 1998. Digital image processing and the recording of rock art, (paper presented at IRAC 1998 “Crossing Frontiers”, Portugal).
- DONNAN, E.F. 1998. An Investigation into the Application of Digital Image Processing for the Recording of Rock Art. Unpublished MA dissertation, University of Durham.
- HADINGHAM, E. 1974 Ancient Carvings in Britain: A Mystery. Garnstone Press.
- HEDGES, J.D. 1986. The Carved Rocks on Rombalds Moor. Ilkley Archaeology Group.
- MARK, R. & NEWMAN, E. 1994. Management of petroglyph rubbing at two pacific northwest coast sites, “American Indian Rock Art” v. 20:19-24.
- MITCHELL, W.J. 1992. The Reconfigured Eye. MIT Press.
- MORRIS, R.W.B. 1970-1. The petroglyphs at Achnabreck, Argyll, “PSAS” 1970-1:33-56.
- TATE, G. 1865. The Ancient British Sculptured Rocks of Northumberland and the Eastern Borders. Hunter Blair.
back to index TRACCE no.