The NBII’s Digital Image
Library – Aggregating
diverse images means
QC and standards
TDWG, October 24th 2008
Annette Olson, PhD
Biodiversity Scientist,
United States Geological Survey
Grapevine Beetle (Pelidnota punctata)
© 2008 Sam Houston
Definition Biological Media
• Media, in this case, meaning audio and
visual media = Moving Image, Still Image,
Sound.
Media, multimedia, and audiovisual all have
connotations that don’t completely overlap
with this definition above.
NBII
The National Biological Information Infrastructure
www.nbii.gov
Initial History of the DIL
• An in-house library in 2002.
• Went online in 2004, with 200 images. All public domain.
• 2004 - partners saw it as a platform to serve, showcase and
store their images. Two partners began putting their
images in. Copyrighted, but allowing nonprofit use.
• In 2005 had 1000 images in, offered 100,000 images.
Broad mission – to serve diverse, quality-controlled
biological images.
Audience and Objectives
• Users – natural resource managers, researchers,
and decision-makers.
• Subject scope is defined by what partners deem
useful (resource managers = species and ecology)
• Public Domain, or Copyrighted with statement or C.
Commons license allowing most nonprofit uses.
(no fully copyrighted images allowed)
• Serve as a repository, but also be a public gateway
to other credible biological image galleries.
http://images.nbii.gov
Jaguar Track in
Pantanal, Andrea
Grosse and John
Mosesso
Amerafrican House Gecko
(Hemidactylus mabouia) - whole body
dorsal view. ©2004 Yuri Huta/Finding
Species
Rufous-collared
sparrow nest
(Zonotrichia capensis),
© 2005 Guyra
Paraguay
John J. Mosesso,
Fish ladder
Giant armadillo (Priodontes maximus),
© 2005 Guyra Paraguay
Japanese stilt grass (Microstegium
vimineum) invasion, © 2004
Elizabeth A. Sellers.
Media as Biological Records
Each image is dynamically
linked to metadata.
Required:
Photographer,
Description
Keywords,
Cataloger
Copyright
Scientific and common
names, when organism
shown. Based on ITIS.
Strongly encouraged.
date
geographic location
physical description,
PRE-SET METADATA
WILL CREATE METADATA
Discover
Life
NatureServe
ReGAP
Morphbank
Finding
Species NPN
FRAMES
US FWS
Smithsonian
Dept. Botany
USGS Personnel
Nature
USGS Projects
Photographer
Indiv. Biologist
Hobby
Photographer
WDIN
NBII
Personnel
SAIN
USGS
Projects
DIL
+ STANDARDS
+ QC
Portlets, RSS
DISCOVER LIFE
GBIF
EoL
GAPServe
Andrea Grosse, Wasp Nest in
Gallery Forest, Paraguay
Solutions
Original images
Mention in metadata
Automatic tools, link checks,…
Feedback mechanisms
Embedding metadata
Signed statement of authenticity and ownership –
images on sets
Image resizing tool
Guiding documents
Metadata templates
QA/QC procedures
Devoting staff to screening, tracking
Issues
• Quality
• Digital editing (restrictions, original, info in metadata)
• Artificial settings (info in metadata)
• Species identification (feedback)
• Metadata – burden of creation
• (progress being made here - extraction EXIF info, pattern
searches…, automatic validation tools)
• Change in the resource over time
• Broken links (auto validation tools)
• Name changes (auto validation tools)
• Technical aspects of linking
• Copyright/privacy/publicity - global
• Standards – or lack of…
Solutions
•
Watermarks – not lost, hard to
alter
•
•
•
Limited in information,
Info changes
Watermarks can potentially cover
information useful for later research
•
Embedding metadata via Adobe
XMP, EXIF, JPEG2000…– can be
stripped.
•
GUIDS (LSIDs) ….. Do not give
information immediately, but
resolvable
Metadata issues
• No one standard fits – digital resource, original, subject




Dublin Core – most common among image galleries, doesn’t
handle biological info…
Darwin Core – specimen oriented
EML – dataset oriented (potential, but media module still needs
to be created)
MODS, PREMIS, DIG35, NISO, “EXIF,” FGDC Biological Profile,
OGC, …
• Encoding Schemes/data values
Multiple subjects within an image
• Describing relationships
• Repeatability - Multiple
species, subjects, time
periods…
• Pairing of fields – species
names with authorities,
common names, and taxon
levels.
Red-crested and Yellow-billed
cardinals (Parcaria coronata,
Parcaria capitata),
©2004 Guyra Paraguay
© 2008 Bruce Avera
Hunter, Red-eared
Sliders on
American alligator's
back
© Ronald Hoff,
White-eared
Puffbird
(Nystalus
chacuru)
Related images, issues
• Different resolutions for display,
dissemination
• Rapid Sequences (time lapse)
• Related images of same specimen
– and the habitat where they were
trapped.
• Related via other subject matter
(habitat changes at a location over
time.)
• Images derived from other
images… (montages)
• Illustrations published in a book
(dc:Source)
• Images linked to a dataset,
specimen
California broomrape (Orobanche
californica), © 2007 Ted
Niehaus/Smithsonian
Metadata – other issues standards
don’t quite cover
• History – image derived from image that is
derived from video…
• Precision of geolocations,
• Methods,
• Language of metadata
• Listing which authority behind a classification
Scientific name source (Catalogue of Life, a
published paper, Sibley’s Guide to Birds….)
Habitat, agriculture (Bailey’s Ecoregions)
• Others…
Many Communities asking for:
• Best Practices
• Quality control
• Process
• Schema
• Flat, simple Definitions docs
• XML
Ruby-crowned Kinglet (Regulus calendula) © 2006 Charles H. Warren
Why another image gallery and not
Flickr, others?

Always more images than there will be galleries to
serve.

Mirroring doesn’t hurt -- preservation.

Capacity building.



Mandate to help serve US Dept. of Interior Images,
includes long-term preservation.
Add value with controlled vocabularies, quality
control, standards, and interoperability.
Treat the images as biological records.
Copyright Issues…
• Global library – learning global copyright law.
• Tied to “Publication.” If disseminating online –
publishing. giving presentations for
dissemination on a website can be considered
a form of publication. For us, probably fair use,
but….
• Images travel!
• Downloaded from websites
• Can be selected from PDFs
• Presentations – slides can be copied
People Issues
1. Right to Privacy – name or even just
image not used without permission, in
some cases even if taken in a public
space.
2. Right to not have their images/face
used for any commercial use/publicity
without their permission. Even if taken
in public space and public domain.
Signed waivers, or statements in
prominent places online.
Photos by John J. Mosesso,
Bird Banding
Facial recognition software is
changing this environment. It’s not just
about names. Especially for kids!!
Descargar

The NBII's Digital Image Library