|
A
Multispectral Critical Edition
|
|
|
|
|
Spectral imaging, when
used for the study of cultural objects, involves illuminating an object
with successive wavelengths of light – starting with
ultraviolet, working through the visible spectrum, and concluding with
infrared – then digitally photographing the resulting
illuminations. The produced images all have exactly the same pixel
dimensions, and are spatially registered, the content being equivalent
at each pixel location in each image. Later, imaging scientists process
this data in order to enhance select features of the cultural object. |
Visit the related
section
from Livingstone’s
Letter from Bambarre.
 |
 |
 |
 |
 |
 |
Figures
1, 2, 3, 4, 5, 6. A
page of Livingstone's 1871 Field Diary
under six of the twelve illuminations used. |
|
The Livingstone team
spectrally imaged the 1871 Field Diary in the Reprographics Studio at
the National Library of Scotland (NLS). The NLS could only allocate
about half of the Studio to the team due to the need to support other
ongoing activities. As a result, the team created a dark room by
purchasing window curtain "blackout" lining at a local store and then
hanging this curtain over the opening between the two halves of the
Reprographics Studio, an expedient that worked surprisingly well. The
NLS also provided the copy stand upon which all Livingstone folia were
placed for imaging, tripods for the lighting system, a computer monitor
for the camera system, and two workstations for the individuals
operating the camera system and data logger.
 |
Figure
7.
Members of the Livingstone team (from left, Boydston, Toth,
Christens-Barry, Simpson) moments before initial imaging. The edge of
the copy-stand is just visible in the foreground left. The "blackout"
lining
curtain, partly pulled up, appears in the background.
|
|
The Livingstone team
transported the EurekaVision™
spectral
imaging system
from the U.S. and
set up the system during the first two days of the imaging sessions.
The EurekaVision™ system, a state-of-the-art image capture
system, maximizes the information recorded with minimum impact on the
historical documents. MegaVision’s PhotoShoot™, the software that powers the system,
is designed to ease user tasks such as image sequence setup, imaging
parameter entry, focus and image inspection, collection and
organization of metadata, and image conversion and storage.
 |
 |
Figures
8, 9. Left: Wisnicki
and Harrison discuss the 1871 Field Diary.
The MegaVision E6
39 megapixel camera hangs overheard.
Right: Karen Carruthers (Property Manager, DLC), foreground, examines
a leaf of the manuscript; Boydston operates the workstation in the
background. |
|
The
components transported by the team to Edinburgh included: |
1) monochrome
MegaVision E6
39 megapixel camera back (7216 X 5412 pixels; 16-bit data with
approximately 12 bits of dynamic range) mounted in a technical view
camera with a 60mm UV-VIS-IR lens and a color filter wheel installed
for UV fluorescence studies. |
2) light-emitting
diode
(LED) illumination panels which were placed on tripods on either side
of the copy stand in symmetric downward-looking positions angled
approximately 45º on either side. The team also used diffusers
to create uniform and nondirectional light fields. Raking illumination
was provided in two spectral bands from either side of the manuscript. |
3) modified desktop
computer that controlled the imaging system using
MegaVision’s PhotoShoot™ 4.0 software. This software
orchestrated the operation of both camera and lighting through FireWire
and USB connections, respectively. |
 |
 |
Figures
10, 11. The LED
illumination panels. |
|
Each LED panel
contains
seven banks of LEDs that emit in the ultraviolet (1 wavelength) and
visible regions (6 wavelengths) of the spectrum and five additional
clusters of LEDs that emit in the infrared region (5 wavelengths). The
geometry of the LED light panels is designed to provide maximum
uniformity of the illumination at the surface of a leaf being imaged. A
primary advantage of the system is that LEDs do not generate heat that
can damage fragile pages. In addition, illumination from a low
“raking” angle enhances the visibility of surface
height variation, yielding images in which impressions or surface
relief features are enhanced. |
The
EurekaVision™ system illuminated each Livingstone folio with
twelve bands of wavelengths of light from the ultraviolet (UV) through
the visible and the infrared (IR): |
Band
= Wavelength in nanometers |
01
= 365 (UV) |
07
= 638 |
02
= 450 |
08
= 700
(near IR) |
03
= 465 |
09
= 735
(near IR) |
04
= 505 |
10
= 780
(near IR) |
05
= 535 |
11
= 850
(near IR) |
06
= 592 |
12
= 940 (near IR) |
|
The monochrome camera
automatically photographed the folio under each illumination, with the
raw image data stored on an external hard drive attached to the desktop
computer. In addition, the team used four raking lights (lights that
illuminate the folia at an oblique angle and so highlight page
topography) mounted within separate panels and on separate tripods when
imaging folia that had not been laminated, as follows: |
Band
= Wavelength in nanometers |
13
= 940RR (IR raking right) |
14
= 465RR (blue raking right) |
15
= 940RL (IR raking left) |
16
= 465RL (blue raking left) |
|
The team chose to use
a
monochrome camera, without color or infrared filters, because doing so
allows: 1) lower light levels for illumination, 2) no interpolation or
preprocessing of the filtered image, minimizing collection artifacts,
and 3) unfiltered light from all spectral bands to reach the camera
sensor. Such a camera yields very pure, high-resolution, registered
images of each spectral band and facilitates processing not only for
text enhancement, but also for very accurate color reproduction. |
The camera operator
and
the data logger collaborated in running the EurekaVision™
system, while a representative of the NLS or the NTS handled the
primary documents and set and flipped them on the copy stand. A variety
of locally-recruited personnel took part in the camera operation and
data logging, including Anne Martin (DLC), Karen Carruthers (DLC),
Suzanne Lamb (NTS), Alison Metcalfe (NLS), and Kate Simpson (Edinburgh
Napier University). Several Livingstone team members also actively
participated. The process required constant rotation of personnel due
to the long hours, continuous concentration required, and complexity of
the data to be logged.
 |
 |
Figures
12, 13. Left: Toth
trains Simpson in data logging.
Right: Simpson trains Anne Martin (Archivist, DLC). |
|
To ensure the team captured accurate metadata about the objects and images, Data Manager Emery developed the Multispectral
Imaging
Metadata
Machine, or MSIMM
(pronounced
"Miss M."), which powered data logging. As each document came up for imaging,
the data logger located the document in the MSIMM database and read off
the file name to the system operator who entered it into MegaVision’s PhotoShoot™.
The data logger, in turn, noted any unusual features of the document or
exceptional imaging circumstances (such as a reshoot due to previous
interruptions) in the log. Once data had been entered, all personnel
put on UV protective glasses, turned off the room lights, and the
capture of the image set began. Each capture lasted about two minutes,
with times slightly extended when raking lights were used. During
capture no one could pass through the dark room because the data could
be compromised and because individuals without appropriate eye
protection could be injured by UV light.
 |
Figure
14. During a break from logging,
Suzanne Lamb (Conservator, NTS) and Karen Carruthers (Property Manager, DLC) examine an
unmarked copy of the 24 November 1869 issue of The Standard held by the NLS.
|
|
The desktop computer powering
the
imaging system also collected the raw image data to an external hard
drive. For each image, the system produced an image set of either
twelve registered images or sixteen if the raking lights were used.
Periodically, team members replaced the hard drive containing the
raw data and took it away for copying, a task which constituted a
full-time job in itself. Team members copied the raw data to other
external hard drives brought by the team and transferred files
externally via FTP overnight to a designated site. This distribution strategy helped prevent the loss of data and made the data available to scientists onsite (Easton, Christens-Barry) and overseas in Hawaii (Knox) for validation, assessment, and initial processing. The strategy also enabled Emery to review the data and metadata for quality and provide updates to the metadata records and software. As a result, the David
Livingstone Spectral Imaging
Project was able to operate twenty-four
hours a day during this period. |
|
|
|