Image Lab

Head: Prof. Dr. Martin Langner
Team: Firmin Forster, Marta Kipke, Maja Leone, Max Maletzki, Michael Schlee

The Image Lab investigates pictures and images in their digital form. It was founded in 2019 as part of the MWK initiative "Humanities and Cultural Studies - Digital". Central research topics of the Image Lab concern formalisation and classification problems of visual testimonies of the past as well as fundamental questions of the scientific and social handling of the digital image. In addition to form-analytical methods of image science, methods of computer vision and image analysis are used. The Lab cooperates with the Information Systems and Machine Learning Lab at the University of Hildesheim and the Corpus Vasorum Antiquorum of the Bavarian Academy of Sciences and Humanities.

Digital Image Science is a field in the Digital Humanities that strives to analyse two-dimensional images using computing methods and applications. Being a Humanity at its core, the images examined often derive from cultural and historical contexts; from art pieces to everday life. We use digital methods in many steps of our research: The acquisition, documentation and processing of image-related data, but also in structuring the images and their metadata in databases. All these steps lead to a better understanding of the pictorial structure, composition and iconography of the images as well as their interpictoriality, iconology and reception in past and present societies.

The field of computer aided image analysis is quickly developing and we are improving and adapting those methods with our oftentimes highly complex and heterogenous data sets. Our research questions regarding the production of the image, their use, their relationship to other images and the people involved in those processes challenge the existing methods to evolve even further. More so than other Humanities, our research is focused on the formal structure of the images, as those properties are the ones most effectively analysed using current computing methods. Especially with Computer Vision and Image Pattern Recognition reaching the Humanities there is the need to translate the complex data surrounding the formal strucuture and the content of the images in a way, that an artificial neural network can prove to be useful in their classification and examination.

The main focus always remains on the variability and diversity of cultural and artistic expressions, as well as the processes and practices related to them. However, instead of concentrating on the already popular and extensively researched "masterpieces", we consider a broad spectrum of images, both in a qualitative approach as well as in their often heterogenous and fuzzy entirety. Especially image pattern recognition harbours great potential not only for the analysis of the images themselves, but also in a historical dimension. Traditional methods of source criticism and context research are still important aspects to apply in this scenario. Our research is the connecting link between a close viewing of the images, as practiced in fields such as archaeology or art history and a distant viewing approach, which is pre-dominantly fueled by computer science. Our case studies accompany the technical development and combine it with traditional methods, being mutually fruitful on both sides.

An example of this is our project EGRAPHSEN, where we train a neural network to recognise ancient Greek vases by their painter. They only sometimes signed their works, so it is a highly sophisticated process to recognise other paintings by a known artist. We study phenomena of „personal style“ by formalising the similarity between images and finding strategies of annotating and processing image data in a way, that a computer aided analysis can be fruitful. With this, a revision and improvement of traditional methods is one of the methodological aspects in our lab.

The critical and reflective handling of visual phenomena and their fluidity must be examined again and again. And last but not least, the theory of the digital image and the manifestations of the digital turn are also the subject of our current research discussion.

In summary, Digital Image Science is preparing, structuring and analysing two-dimensional representations of cultural heritage with computer aided methods. The presentation of the data and the reflection on the methods are part of our work as well. However, the digital image in itself is a phenomenon with scientific and societal implications, which has to be considered at every step of the way.

You can find more information about the Lab's activities here.