Last week, we were very lucky to be visited by two internationally renowned experts in multispectral image processing, Keith Knox and Ivan Shevchuk. We have been eagerly looking forward to this event since 2020, when their plans to present at the More than Meets the Eye conference were postponed by the beginning of the pandemic. Now two years later, their visit offered a wonderful opportunity for the IISICCA team and other interested members of our community to get hands-on experience in multispectral image processing software from experts and developers in the field.
For their presentation, we gathered at the Iowa Institute for Biomedical Imaging (IIBI)’s visualization lab. Among the participants were members of the IISICCA research group, the conservation team at the UI Libraries, as well as faculty and students from several departments across the UI campus. True to the spirit of learning and community, this two-day event— part presentation, part demo, part hands-on workshop—was carefully developed to make the topics of image processing and multispectral imaging analysis engaging and communicable to all in the audience, regardless of their prior experience in the field.
Starting the day, Ivan Shevchuk, who works as an imaging specialist at the University of Hamburg’s Centre for the Study of Manuscript Cultures, gave an overview of the types of materials and inks commonly found in the historical manuscripts he studies. His presentation demystified some of the terminology used by experts in the field, offering a valuable introduction to multispectral imaging and image processing software for those in the audience encountering the topic for the first time. Among the many topics covered was the important notion of a “palimpsest”—that is, a document that contains multiple layers of writing, with earlier layers having been effaced to be used again as a blank writing surface. On parchment, this process of effacement is done through a kind of erasure, where the topmost surface of the animal skin is literally scraped away, leaving only traces of the original ink, that can be all but invisible to the naked eye. By imaging these and other types of damaged documents under exposure to different wavelengths of light, invisible layers can be recovered and even read by the trained eye, revealing a text that would otherwise have been lost.
Image processing plays a crucial part in this process, by allowing the investigator to highlight certain regions or layers of a text that may be visible under some wavelengths of light and hidden at others. Through this process, scholars and conservators can gain valuable information as to a manuscript document’s past life, whether that be knowledge into the text that parchment originally made up a part, or insight into the underdrawings that a scribe or illustrator might use at an early stage in their process of manuscript decoration.
Leading part two of that day’s event, Keith Knox, who is Chief Science Advisor for the Early Manuscripts Electronic Library (EMEL), walked participants through the first steps of viewing and processing multispectral image data in Hoku, a general-use, Java-based software application Keith developed for processing multi-spectral image data of damaged or erased manuscripts. Workshop participants were given the chance to explore the program on their personal computers with a sample dataset Keith and Ivan prepared in advance: a leaf from a musical manuscript that contained multiple layers of writing: not only the musical notation and Latin writing visible to the eye, but also at least two underlying layers of text left behind by previous scribes in an earlier version of the text (or even, in one case of mysteriously reversed lettering, perhaps the accidental transfer of wet ink from another document altogether!).
On day two of the week’s workshops, participants were guided through the process of using ENVI, an image processing and analysis application that can be applied to a wide range of data, from multispectral imaging of manuscripts to geospatial satellite images. Like Hoku, ENVI allows the investigator to set specific visualization parameters based on the particular feature of the manuscript sample that they want to focus in on—for instance, a layer of underwriting or underdrawing that lies beneath the primary text—and apply those processing settings to a large batch of data, reducing the time and labor required for the analysis of a full manuscript or book.
Throughout the workshops, both Ivan and Keith generously offered their assistance with questions both technical and theoretical, engaging with both the imaging experts in the room and the newbies like myself. Learning new concepts and software applications isn’t always a walk in the park, but Ivan and Keith made it a truly unintimidating introduction for all of us. Thank you, Ivan and Keith, and looking forward to many future connections and collaborations!