Recovery of Damaged Writing
In the work on recovering damaged writing, imaging hardware has made huge progress in the last decade. Besides multispectral imaging that uses imaging sensors with high spatial resolution (many pixels) and new, powerful narrowband LEDs at 19 (or more) wavelengths for illumination in the UV, visible and NIR portions of the electromagnetic spectrum, many different systems for spectral imaging, such as hyperspectral and XRF, have been designed and successfully applied. All these systems possess quality control operation charts and can deliver high-quality image data. We expect that in the next 5 years further commercial multispectral imaging systems optimized for easy set up and operation will appear on the market. Yet, capturing images constitutes only the first step towards the recovery of damaged writing. The second step consists of image processing, that is, the selection of the most appropriate methods of statistical analysis to enhance the contrast in the raw images and visualize the text. Unfortunately, in terms of automated procedures and quality control operations, the state of the art of automated statistical image processing is far behind that of the hardware for image capture.
In the past decade a single software package, ENVI (current version 5.5) of Harris Geospatial has emerged as a primary tool for statistical image processing of multispectral data of objects of cultural heritage. ENVI was initially developed as an imaging tool to enhance details in imagery from the Mariner probes to Mars in the 1970s. It has since been developed into a market leader across remote sensing scientific and military fields and applications that rely on satellite and aerial spectral information capture and feature extraction using statistical methods. ENVI is also widely used in the field of cultural heritage imaging, including 44% of the European community in this field in 2017.
Besides ENVI, a new software tool for treating and extracting features from multi-, hyperspectral and X-Ray Fluorescence image data is being developed. This independent HOKU software has a graphical user interface (GUI) allowing the automatization of statistical processing steps as well as basic image treatment for multiple datasets. Though less powerful than ENVI, it is free of charge and user-friendly, which makes it highly attractive to scholars and scientists alike. This project aims at further development of the processing tools in collaboration with the project RFA05 as well as external partners from the Cambridge University Library, Rochester Institute of Technology and the Early Manuscripts Electronic Library.