Everyday Life in the Middle Ages and Digital Image Analysis

Gerhard Jaritz (Krems an der Donau/Budapest)

History of everyday life is to a large extent a research into patterns that is to be done in a comparative way. It has to connect the quantitative with the qualitative to be able to do proper analysis of the repetitive and of the routine of life (Jaritz 1989, Jaritz 1994). One has become aware of the fact that individual examples are on the one hand necessary, on the other hand only bricks and pieces for a more pattern-lead research and their results. The role that pictorial sources play in this research can be seen as very relevant. In this context, we are certainly aware of all the problems arising with the "reality" or "reality effect" of the contents of images (Moxey 1994, Moxey 1995, Jaritz 1995); but they should not be discussed here.

During the last few years, the importance of integrating image information into art historical (Hamber, Miles And Vaughn 1989) and historical research, and particularly into the history of everyday life, has been increasing decisively. For a long time images had been playing a more or less marginal role as accepted sources in most fields of conventional historical research. They had been a domain of art history. In history, they only got well acknowledged as a medium to illustrate some text and/or to be verbally described in connection with this text, but were only rarely considered to be able to serve as a proper means for systematic historical analysis comparable to the one of written sources. Due to the above mentioned changes and developments, pictorial sources have started to loose that function as mere illustrations and have reached a similar level like written and archaeological evidence. An important role in this process is played by the application of methods of digital image processing (Thaller 1992). Particularly, databases including the connection between standardised verbal description of the images and the digitised image information itself prove rather relevant and helpful. That way, especially the research into the history of everyday life and of material culture have been starting to concentrate on the analysis of (patterns of) messages borne by images and to put them into broader contexts.

Working with images in the mentioned field, the method to be applied is at least at the beginning more or less similar. Every image to be included into a documentation, a research project, an educational package or a broader systematic analysis of visual sources has to be described. It must be stated, though, that - except of upper level description fields like artist, theme of the image, dates, provenance etc. - the general strategy (e. g., full text description vs. standardised term description; different classification systems; the depth of descriptions, etc.), the terminology and the hierarchy of the descriptions are rather different and underlie standardisations often only to a certain degree. Just recently, however, various but again quite different and not yet generally excepted efforts have been initiated to provide acknowledgable standards, particularly concerning classification and terminology (e. g., ICONCLASS, Heusinger 1989; Garnier 1984, Groupe 1993).

Those differences are certainly dependent on the diverse applications and usages of image databases. An even more problematic aspect, particularly with regard to systematic iconographical analysis of visual sources, is the depth of the description of images, which is dealt with very differently and not following any accepted standards. The same often proves true for the relations and hierarchies of objects depicted in images. The major differences with regard to those problems are at least to some degree caused by the size of image collections. It is certainly another situation to describe images of a certain topic and in a restricted number for some specialised research project, where you are able and more or less have to go into every detail, than to be confronted with a huge collection of visual sources of some hundredthousands aiming to provide general information for many aspects of iconographical research (Cassidy 1991). In the latter case it seems to be more or less impossible to provide a systematic, detailed description as it is necessary and feasible in the first case.

This paper is going to deal with some possibilities of such research that are mainly based on medieval and early modern source material - especially panel paintings, frescoes and manuscript illuminations from Central Europe -, studies which are done at the "Institut f\"ur Realienkunde des Mittelalters und der fr\"uhen Neuzeit" of the Austrian Academy of Arts and Sciences, an institute concentrating on the research into the everyday life and the material culture of the Middle Ages and the early modern period. Particularly the application and the usage of "kleio" and "kleio IAS" (Image Analysis System), a public domain database management system that has been developed at the "Max-Planck-Institut f\"ur Geschichte" at G\"ottingen (Germany) (Thaller 1993, Woollard and Denley 1993, Jaritz 1993a), can be seen as a decisive step towards standardised and systematic analysis. It is used in very different projects, in which image analysis gets necessary. At the "Institut f\"ur Realienkunde" it is applied under UNIX for the database REAL, a public domain image database for the research into medieval everyday life and material culture.

The mentioned image-text-context can be seen as an indispensable means for comparative studies of various images or their details respectively and their verbal descriptions. Though such comparisons concerning the used image material of medieval paintings still mainly have to be done visually and not yet based on automatic pattern comparison or pattern recognition, they nevertheless offer various possibilities for the research into certain components of depicted situations, persons, objects, "qualities" and their visual comparisons by the historian.

Beside a structured description of an image and of details and parts to be recognised in this image - object of art, documentation, commentary, persons, their attributes and parts, objects, their parts and parts of parts, and diverse relations to them -, there certainly is more or less often the wish or need of a kind of full-text description of an image, at least in the way of a "regestum", giving on the one hand full-text information of the contents of the image, on the other hand allowing the application of certain possibilities of full-text analysis.

Original texts play an important role as part of medieval images. Stories are depicted and written down. There it is not only necessary to transcribe the texts in the course of the description of the source, but also to keep those texts as cutout segments and bound details. Those text cutouts certainly can be treated by zooms, by filter, transformation and other image enhancement routines to make them better readable - with the same routines which we are using when working with digitised written original sources like charters or manuscript pages.

As a pre-condition, possibilities of a useful digital image processing system must exceed those ordinary routines offered by most of the commercial systems. This implies, e. g.,