VASARI, MARC, ACOHIR and Viseum

1
VASARI, MARC, ACOHIR and Viseum

• Advanced imaging systems
• Kirk Martinez
• Dept. of Electronics and Computer Science
• University of Southampton

2
VASARIVisual Arts System for Archiving and
Retrieval of Images

• European project - 1989
• Aimed to make highest quality imaging system to
  replace conventional photography
• BRAMEUR LTD, DOERNER INSTITUT, TELECOM PARIS,
  NATIONAL GALLERY, THOMSON, BIRKBECK COLLEGE, TÜV,
  GFAI, SIDAC, SYSECA, DIRECTION DES MUSEES DE
  FRANCE

3
National Gallery VASARI Scanner
Captures 3kx2k patches precise mechanics uses
Kontron Progres camera
4
VASARI scanner
Fibre-optic system illuminates area with narrow
spectral band of light
5
Vasari lighting system
6
VASARI scanner filter wheel
Filter wheel in light system holds the
narrow-band filters recently extended from 7 to
12 to include near infra-red bands
7
VASARI scanner in action
8
CIELab for device independence
9
The Uffizis VASARI scanner
10
(No Transcript)
11
From VASARI to MARC
12
MARC

• European project to investigate accurate printing
  and a calibrated image chain.
• calibrated printing press and printed a book!
• Also made a new camera with the Lenz brothers in
  Munich.

13
MARC camera
14
Sensor Masking
15
(No Transcript)
16
(No Transcript)
17
MARC results

• 20min capture time for 10kx10k
• in use at National Gallery photographic dept. to
  capture the whole collection.
• colour accuracy around 2-3?E CIELab

18
MARC2 camera

• Made by the Lenz brothers in Munich for JJT and
  National Library of Congress
• much faster around 30s!
• 10kx10k x 12 bits
• uses a 1.3kx1k masked sensor in 4 macro positions
  and microscanning
• uses fibre link to PC with 1GB RAM.

19
Viseum

• How to browse 2GB images over the Web?
• Web server plugin sends image tiles
• Java applet views portions of images at any scale
  and position

We also made a fibre linked network from
Vancouver to London to test ATM (up to 155Mbit/s)
20
(No Transcript)
21
ACOHIR Accurate colour high resolution imaging of
3D objects

• Improving on Quicktime-VR
• Higher resolution
• more accurate colour
• Web browsing of objects without large download
• Make applications for archaeology, sculpture and
  commercial sales or porcelain and furniture
• ran from 1997-1999

22
ACOHIR Partners

• Athens Technology (Gr)
• Barco (B)
• University of Southampton
• ENST, Paris
• AIDIMA (E)
• Cobax (D)

• The National Gallery, London
• LRMF - Louvre (F)
• Museum of Cycladic art (Gr)
• University of Thessaloniki (Gr)
• Christian and Byzantine museum of Athens
• EOCENO (E)
• Vasari enterprises (UK)

23
ECS Imaging System
24
Louvres ACOHIR system

• Uses Kodak DCS 560 camera (3kx2k)
• strong turntable for larger sculptures
• custom PC software for capture and colour
  calibration.

25
Acquisition Procedure

• Setup object in centre of turntable.
• Focus camera (using zone plate print if needed).
• Set aperture and gain for Macbeth white to be in
  range.
• Take test grab to check exposure, focus and
  cropping.
• Take black image (cover lens).
• Take image of flat white where chart will be.
• Take image of Macbeth chart.
• Run capture script.

26
Calibration Procedure

• Chart Linearize, subtract dark, scale by white,
  measure
• Use MacBeth image to find 3 by 3 matrix, plus 3
  offsets, to go from linear RGB to XYZ
• Colour convert each frame to XYZ, then to sRGB
• Optional sharpen
• Save as TIFF pyramids

27
Tiled Pyramidal JPEG TIFF
Store full image plus half/quarter etc
resolution Each image is tiled as 64x64
blocks Each tile is JPEG compressed (or LZW
lossless) Pels are stored as sRGB or CIELab
28
Icon View Main image area Zoom controls
Description links to views off-site links
29
Current research

• content-based retrieval of art
• finding images for the VA picture library
• finding conservation structures for Louvre and
  Uffizi
• Demos www.ecs.soton.ac.uk/km
• email km_at_ecs.soton.ac.uk