Se ha denunciado esta presentación.
Utilizamos tu perfil de LinkedIn y tus datos de actividad para personalizar los anuncios y mostrarte publicidad más relevante. Puedes cambiar tus preferencias de publicidad en cualquier momento.

Ki presv2

324 visualizaciones

Publicado el

  • Inicia sesión para ver los comentarios

  • Sé el primero en recomendar esto

Ki presv2

  1. 1. TSB / MEDIA ProjectImage Processing for digital stores and Usage tracking
  2. 2. Problem● Cultural Institutions have rich collections of digital assets, but there are many unknowns around opening up these assets for public consumption.● There is a great deal of [conflicting] pressure on these institutions to (a) Openly share both metadata and resources and (b) be come financially self sustaining
  3. 3. MEDIA● Supports institutions in making informed choices and supporting their chosen strategy● From Selling images (Soutron Side)● Through to watermarking, steganogoraphy and provenance tracking
  4. 4. Vision● Cloud hosted secured image repository that can be configured against a host collection management system after a CC “Paywall”● Zero local install● Ability to manage multiple collections and systems, with differing rules and processes
  5. 5. Note● Following screen captures are a loose interface on top of MEDIA services● Our focus has been on configurable parameter driven adapters that can harvest metadata into a canonical form for image processing● The screens represent a [rough] attempt to visualise how a product might look.
  6. 6. Accounts create contextsDefined by URLHere – default “admin” context
  7. 7. Contexts contain stores, root of URLsHere /knowledgeintegration context
  8. 8. Ingest● The process of synchronizing the media repository with a host institutions collections and digital assets.● Extensible system of “Connectors” for extracting image resources and metadata from a wide range of collection management systems
  9. 9. Stores contain resources,Define ConnectorsAllow resource upload
  10. 10. Add connector – Wire this store upTo a collections management systemOr other source
  11. 11. Ingest Processing● Majority of our work here – Flexible (DSL) driven tool for – Image format translation – Image Metadata Generation & Embedding – Steg Hide Identifiers in Document & Metadata – Create Variants: Resize, Watermarks, Add Metadata, Create Thumbs [& Pyramidal Tiffs] – Record locally and Index for Store –
  12. 12. Stores contain images,Configure rules,Define connectors,Allow individual upload(Watermarked rezisedthumbs)
  13. 13. View original full size - But with full(Configurable) visible watermark
  14. 14. Result of “Purchase” link:[Really new copy function] - New Copy “Item” [FRBR] – Newly minted embedded item copy code [GUID] No Visible WatermarkCopy/Item specific XMP and EXIF metadata for item Steg hidden ID of new item Image is unique to this occurrence of a purchase, Can be linked back to purchaser Embedded metadata includes docid – OneOf our toys is an image spider - but Google Search?
  15. 15. Use steg information to find original “Buyer”
  16. 16. Doc owner tool - System uses steg to extract copied doc IDand tells us who copied/purchased it originally
  17. 17. Unified SearchingSurfacing MEDIA functionality in end user applications
  18. 18. Because properties like price are now user dependent Physical union catalogues cannot be used hereThis tool uses a map-reduce approach to x-search, collating results from several sources and making use Of web-scale clustered architectures. Old style x-search did not scale well to cloud infrastructure (Session pinning problem)
  19. 19. Result from soutron – Price per user, Metadata profile for purchase info NCMG Result, no buy But image secured, vis WM, Steg API At work – Facets, Sort, Cache
  20. 20. Other toys● Spider that finds copies of images – Even better when google/other engines support EXIF metadata search – Still useful concept – Report on all copies of an image