This item has the following downloads:

( PPT )

( PDF )

RESEARCH TOPICS


* Web-Interface Performance

* DTD Extensibility

* Imaging

* Distillation


m Other topics?


ABCDl
EFGH
, IJKL
MNOP
SQRST
!UVWX
YZ


2002 September -- ejk/UF








CONTEXT


Image Only Pilots
Australian Periodical Publications, 1840-1845
National Library of New Zealand. Papers Past
Image & Indexing/Tagging Pilot
University of Florida. Caribbean Newspaper Imaging Project
University of Florida. Florida Newspaper Project
Image & OCR Pilots
Lambrakis Press Archives
ProQuest. Historical NewspapersTM
TIDEN Project : a Nordic Digital Newspaper Library
Olive Software Pilot
The British Library


2002 September -- ejk/UF








WEB-INTERFACE PERFORMANCE



Primary Purpose:
Characterize the bias of individuals conducting study


Products:
How to use ActivePaperTM to YourAdvantage
Integration with CONTENTdm, XPAT 5.0, other
Alternate deliverable images
Centralized service Distributed content Variable platforms


2002 September -- ejk/UF








DTD EXTENSIBILITY



Primary Purpose:
Assess the XML against established newspaper uses


Products:
How to use ActivePaperTM to YourAdvantage
Document the XML as a public DTD
Establish a maintenance authority
Provide for extension of the DTD
Automation for extended tagging
How to construct a style sheet
Integration with CONTENTdm, XPAT 5.0, other
Define issues per the Economic Model


2002 September -- ejk/UF








IMAGING



Directory Structure and File Naming
Archival Formats
Optimized Imaging


2002 September -- ejk/UF








IMAGING: Directory Structure and File Naming



Primary Purpose:
Recommended practices


Products:
Methods for dealing with anomalies
Automated name capture during imaging


2002 September -- ejk/UF








IMAGING: Archival Formats


Primary Purpose:
Description of file formats & their characteristics
for archive, distillation, and distribution


Products:
Preservation metadata
Anticipate migration
Schedule & fee structure for inspection
& migration
Strategy for format migrations & emulation


2002 September -- ejk/UF








IMAGING: Optimized Imaging


Primary Purpose:
SBest practices for microfilming and digitizing
(quantitative assessments)
Film reduction ratio
Evenness of illumination on film
Film background density
Quality Index & DPI/PPI
Skew
Color-space & Bit-depth
Image density/black & white points
Despeckling and Sharpening
Image restoration methods


2002 September -- ejk/UF








IMAGING: Optimized Imaging


Environments:
Operating System
Scanning Hardware
Lighting and Light Filtration
Post-processing

Other?


Other Products:
Control target for OCR assessment
Revision: RLG Preservation Microfilming Guidelines


2002 September -- ejk/UF








DISTILLATION


Document Zoning
Optical Character Recognition

ABCD
EFGH1
MNOP
QRIST
iUVWX
YZ


ABCD
EFGH'
IJKL
'MNOP,
QRST
.UVWX
YZ
2002 September -- ejk/UF








DISTILLATION:


Document Zoning


Primary Purpose:


Confirm assumptions re: document zoning
OCR has difficulty processing large letters
Smaller zone yield more accurate text


Products:
Establish reference to the ...
PDF (fully scaled)
TIFF
Other derivative file formats (fully scaled)


2002 September -- ejk/UF


ABCD
EFGH H
IJKL
MN OP1
OQRST
VUVWX
Uy7w
YZ


ABCDI
EFGH'
I J KL
'MNOP
QRST
UVWX
YZ








DISTILLATION: OCR


Primary Purpose:


* Provide quantitative OCR accuracy information


Areas of Investigation:
Distillation Source Images
Language and Fonts
Column & Line Density
Relative Density/Contrast
Text Curvature and Other Defects


2002 September -- ejk/UF


ABCD
EFGH :
IJKL
MNOP
QRST
UVWX



ABCDI
EFGH'
IJKL
'MNOP,
QRST
UVWX
YZ








DISTILLATION:


OCR


Distillation Source Images


ABCD
EFGH:
IJKL
MNOPR
jQRST
UVWX
YZ
i Y-ZJ'4 Ti
i 1~U 'IPT) '


Primary Purpose:
Predict accuracy contingent upon source document
(printing technologies & filming standards)

Test-Set Characterization:
Source type (newspaper or microfilm)
Production date (technologies & standards used)

Additional Products:
Best practices
Accuracy : Cost Matrix


2002 September -- ejk/UF


ABCD1
EFGH'
IJKL
;MNOP1
QRST
!UVWX
YZ








DISTILLATION: OCR
Language and Fonts


Primary Purpose:
Demonstrate ability to distill languages, character sets
& fonts

Test-Set Characterization:
Language & character set groups
Font face & font size groups
Regional variant spellings

Additional Products:
Olive Software Speaks Your Language
How Olive Software Learns Your Lingo
Stylized text recognition & distillation guide


2002 September -- ejk/UF


ABCD
EFGH:
IJKL
MNOPt
QRSTB
UVWX
YZ
-Z '
i 1~U 'IPT) '


ABCD1
EFGH'
IJKL
MNOP,
Q RST
!UVWX
YZ








DISTILLATION: OCR
Column & Line Density


Primary Purpose:
Demonstrate ability to distill compact text

Test-Set Characterization:
Pre-1900 newspapers
Advertisement pages
Pages predominantly 8 pt. type or less
Pages with less than 1 mm space between lines
Pages with characters spaced at or below mm


2002 September -- ejk/UF


ABCD
EFGH:
IJKL
MNOPR
QRSTB
VUVWX
YZ
- -' - -
i 1~U 'IPT) '


ABCD1
EFGH'
IJKL
M NOP,
Q RST
_UVWX
YZ








DISTILLATION: OCR
Relative Density/Contrast


Primary Purpose:
Investigate low and uneven contrast materials

Test-Set Characterization:
Low contrast pages
Pages with low contrast zones
Printing, Filming, & Age/Storage Defects

Additional Products:
Best practices
Accuracy : Cost Matrix
Don't forget to buy the Life Insurance


2002 September -- ejk/UF


ABCD
EFGH:
IJKL
MNOPR
QRSTB
YUVWX
SY -Z' I
i 1~U 'IPT) '


ABCD1
EFGH'
:MNOP,
Q RST
_UVWX
YZ


E








DISTILLATION:


OCR


Text Curvature and Other Defects


Primary Purpose:
Benchmark current capability to distill curved text
& other defects of printing or filming

Test-Set Characterization:
Curved text zones
Broken character zones
Broken line zones
Garbage elements (stains, etc.)

Additional Products:
(Additional automatic image correction processes)


2002 September -- ejk/UF


ABCD
EFGH:
IJK L
MNOPt
QjRST
UVWX
YZ
-ZJ'
i eb,] ~1~4 T i


ABCD1
EFGH'
IJKL
M NOP
Q RST
UVWX
YZ