Multispectral Imaging (MSI), November 2023, Biblioteca Capitolare di Verona

Dataset Description This Multispectral Imaging (MSI) dataset originates from the Palimpsests in Danger project. MSI is applied to palimpsest handwritten codices (manuscripts) from the Biblioteca Capitolare di Verona, which have been previously treated with chemical reagents. This dataset includes multiple high-resolution images taken under varying wavelengths and angles of light with narrowband illumination, high fidelity color RGB images deriving from narrowband visible wavelengths, and processed MSI files, documenting the manuscripts' historical context and surface details. This enables thorough examination and preservation of the object. Imaging Equipment Description/Specifications A monochrome digital camera, a 120mm Apochromat macro lens, a digital electronic shutter, and a computer-controlled filter wheel are used alongside narrow band illumination sources. Two stationary lights are set at a 45-degree angle to illuminate evenly the object to be photographed. Additionally, and for specific sequence of images, a narrowband light source illuminates the object with direct light. Color targets and small reflective calibrated surfaces (Spectralon) are used to calibrate color, white balance images, and calculate light efficiency. The specific wavelengths, combination of filter wheels and angle of the illumination source(s) can be determined by records of descriptive filenames. Camera - MegaVision Q15 Camera featuring a dual stage TE cooled CMOS sensor of 150 Megapixel (SONY IMX411BSI back-illuminated sensor). 14192 x 10640-pixel array with 3.76 um pixels. The sensor size is 54 mm x 40 mm and is able to achieve ultra-low levels of dark noise - 1e to 3e. Native 16-bit A/D - 65536 levels. Full Well - 50ke- >80ke. Remotely controlled by computer and capturing software. This camera is mounted facing directly down onto the target object and calibration targets. Light sources - 2 x MegaVision Spectra Palette lights housing 16 narrow bands (365nm, 400nm, 450nm, 470nm, 505nm, 530nm, 560nm, 590nm, 615nm, 630nm, 655nm, 700nm, 735nm, 780nm, 850nm, 940nm), 2 x MegaVision Auxiliary IR narrowband panels (1050nm), 2 x MegaVision narrowband raking raking panels (450nm, 940nm), 2 x MegaVision Auxiliary UV panels (385nm, 420nm), 1 x MegaVision LightSheet Panel housing 4 narrowband lights (630nm, 735nm, 940nm, 1050nm) Lens - MegaVision 120mm f/4.5 Macro UV-IR Apochromat incorporating both floating element and advanced UV-VIS-IR apochromatic color correction technology. Filter wheels - 2 x MegaVision filter wheels housing 3 filters each for a total of 6 filters (R25, G58, B47, R72, 87C, 665). Filters R72, 87C and 665, allowed us to experiment with the effect of Infrared Fluorescence. Targets - 1 x X-rite Color Checker Classic, 1 x Macbeth color checker mini, 3 x Spectralon patches of 99%, 50%, 5% nominal reflectance. Software - MegaVision Photoshoot v5 image capture and processing software. Photoshoot communicates with all hardware components, allows the use of predefined and highly configurable image sequence capture, handles metadata, file naming as well as flattening and color creation tasks. Data Structure Data is divided into three main categories and main folders. Accurate_color, Flattened, Processed. The Accurate_color folder includes TIF files of color images for each page of manuscript that was imaged with the MSI equipment. These images are a combination of 11 images captured under narrowband visible wavelengths for each page. The Flattened folder lists subfolders for each of the pages captured with the MSI equipment. Each subfolder includes 51 or 67 monochrome TIF files. Each of these files has been captured with a different combination of lights or filters. Folders with 51 images include a basic set of MSI data, while folders with 67 images include the basic set in addition to extra images captured with the object being illuminated only by the light aiming directly perpendicular to its surface. These later set enabled us to experiment with the effect of glossiness and reflections of certain chemical elements, and evaluating the need of Reflectance Transformation Imaging (RTI) techniques. Folder and file names Folder names are composed by the shelfmark (identifier) of the object, the folio/page number, and the description of the technology applied. Example: msXL_319r_MMSI, msXL is the shelfmark, followed by folio 319r, and finally indicating that MMSI technique was used to capture the images of this folder. For some folders the additional “+overhead”, signifies the extra presence of images captured by the illumination of the object from a single light source, place perpendicular to the imaging area. Filenames are composed by the shelfmark (identifier) of the object, folio/page number, technology used for the capturing of this image, followed by a human readable description of the lighting source, specific wavelength, presence of filter (if any), color category of wavelength and a sequential number of the capture order. MB = Main Bank, main source of lights illuminating from a 45-degree angle RL-RR = Raking Left - Raking Right, light sources illuminating from a 15-degree angle W = Wheel, light sources illuminating from a 45-degree angle, in combination with presence of specific filter in front of the lens element. TX = Transmissive, light source is directly underneath the object illuminating evenly and utilizing transmissive light Example: msXL_319r_MMSI-MB365UV_01, msXL is the shelfmark, 319r is the folio number, MMSI is the capturing technology, MB365UV is the 365 UV light illuminating from the 45-degree main bank light sources, 01 signifies that the image is the first of the imaging sequence. Image Processing Description Software Hoku, ENVI, as well as Photoshop are used to calibrate and process MSI data. The most prominent among the techniques described below is combinations of Principal Components Analysis. PCA: Principal Components Analysis ICA: Independent Components Analysis Date of Captured Data November, 2023 Data Captured by Damianos Kasotakis Data Processed by Keith Knox Russell Knox Roger Easton Kyle Huskin