To provide such repository services to the Berkeley campus, Information Systems and Technology, The University Library, and campus museums and archives have collaborated in developing prototype repository services to support digital imaging projects in campus museums, archives, and libraries. The implementation of an image repository will be a primary activity of MIP for the coming year. This repository will provide a testbed for collaboration with instructional and research units at Berkeley in the development of an institutionally supported information technology infrastructure and of institutional policies for management, access, preservation and reuse of digital objects.
During the last year, MIP has moved images from the SMASCH project to an 800 gigabyte robotic tape storage device for image review and organization. This device is insufficient to place all original image files from project to a unified mass storage environment. For networked presentation of images, MIP and Central Computing Services (CCS) have operated a 70 gigabyte optical storage device. This device is insufficient in size for networked delivery of derivative images and cannot be scaled to a larger size. Therefore, further progress toward the establishment of repository services will require significantly larger mass storage devices.
Both the image repository under development by MIP and the course web site management project being developed by ITP, Student Information Systems, and CCS require mass storage. In addition, the establishment of mass storage facilities will facilitate collaboration with the University Library in the investigation of storage, migration, management, and access issues connected with establishing a digital archive for long-term preservation of digital objects. It is estimated that from projects supported by MIP and by the University and Bancroft Library approximately five terabytes of original image files have been created to date.
CCS and MIP have developed plans for the use of a seven terabyte tape robot and the acquisition of approximately 250 gigabytes of RAID storage operated by CCS for image storage and presentation. Funding has been requested through the IST budget process as a priority item in order to undertake the provision of repository services.
IST, the University Library and other relevant groups will establish a capacity planning process to manage the growth of these facilities over the coming years. Mass storage is becoming a campus-wide issue for both academic and administrative data. In this regard, the example of Indiana University in developing a mass storage strategy for its entire campus will be instructive in the development of appropriate strategies for Berkeley in this area.
Access to parts of the image repository may be limited due to licensing restrictions imposed by commercial providers of image data or by campus policy. Therefore, copyright and distribution data are managed as part of the repository. Each derivative image used for public access over networks will contain this information as well as a unique identifier established through a registration authority operated as part of a global registration process under the auspices of ISO.
Authentication services are part of IST Core Programs. Communication and Network Services (CNS) is working to implement Kerberos and X.509 digital certificate servers for the campus. Such services for the image repository will be coordinated with these efforts.
MIP has used the campus Kerberos authentication service to authenticate students in the Department of Architecture who are using the SPIRO database. A Java servelet communicates to both a client and the campus Kerberos authentication service for this purpose. This prototype as well as enhanced Java support for authentication will provide the basis for further work in this area.
Metadata specifications relevant to image databases that facilitate interoperation among library and museum databases have been published by a variety of groups. These include:
MIP is developing a database of these specifications that tracks the specification, the tags it uses, and the definition of these tags. Each tag used will be mapped to table(s) and attributes in each of the databases maintained by MIP. The goal is to generate XML encoded records that identify the specification used, the tags included, and the values for these tags generated dynamically from MIP databases. Therefore, in response to a query based on one or more metadata tags for a particular specification, records from any MIP database that satisfy that query can be generated.
The data structures for this database have been implemented and the building of the database is underway. Populating this database will be coordinated with our data model and data dictionary work described above. Java tools for query processing and record generation will be developed using the application development environment described below.
High-resolution images captured by campus imaging projects cannot be presented on networks without tiling, either static or dynamic. Most current web applications use a combination of JPEG images of approximately 100 dpi and JPEG thumbnails at approximately 15 dpi as two-level tiling of a high-resolution image. This approach does not permit access to the content available in high-resolution images with an original resolution greater than 100 dpi. To address this problem, the ISO JPEG Working Group developed a digital image exchange format, Still Picture Interchange File Format (SPIFF), that describes a method for tiling high resolution images.
MIP is using a commercial implementation of SPIFF, the Jpeg Tiled Image Pyramid (JTIP) format for the conversion of original high-resolution images into a pyramid of tiled images. The JTIP format permits the preservation of the content of the original image and the bandwidth efficient transfer of image tiles over networks.
We are converting high-resolution original images derived from the projects listed in Appendix 1 to this format for networked presentation of these images. During the coming year images managed by MIP will be converted to this format. Discussions are underway with the University Library on the use of this format for images derived from various projects that are supported through Library Systems and Library Photographic Services.
One of the primary uses of these repository services will be to provide management tools to support the campus imaging projects described in Appendix 1. In general, original images are currently housed on tape stored at MIP or, in the case of the University Library, on CD-ROM. It is imperative that these images be moved into an integrated storage environment before their storage media degrade. As the number of images grows, their management on tape or CD-ROM becomes increasingly problematic.
The establishment of this repository will be critical for Berkeley campus participation in the Library of University of California Images (LUCI) project. This project has been organized by slide collection curators at Berkeley, Riverside, and Irvine to provide shared access to digital images for instruction in art practice, art history, architecture and classics. Slide collections from throughout the University of California system are discussing how the LUCI project can be used to coordinate the development of system-wide sharing of digital resources for teaching in architecture and art history.
Most University of California campuses support slide libraries for teaching in art practice, art history, architecture and classics with a significant overlap in the slides curated at each location. To minimize duplication of effort, it is that digitization proceeds in a coordinated manner and that digital images are shared throughout the University of California system. LUCI will be a key project for collaboration with the California Digital Library project.
The Office of Media Services (OMS), ITP, and MIP are collaborating to explore the use of the repository for instructional purposes in art history at Berkeley. Tests are underway to compare analog image projection using standard projectors with JTIP images using the improved network aware projection facilities installed in classrooms at Berkeley. These tests will allow direct evaluation of the utility of digital images for instructional purposes. On the basis of favorable evaluation by faculty and curators of campus slide libraries, faculty will be enlisted to participate in trial use of the repository with the Luna Imaging application, Insight (see Application Development Support Services), for teaching. Depending on funding, these trials can begin during the next academic year.
Both the ECAI project (Appendix 1) and the University Library are interested in establishing the facilities and policies for long-term archiving for the preservation of digital objects. At the present time, these issues are being actively discussed within the library community with no clear direction. The image repository will provide a testbed to investigate and recommend appropriate strategies for the Berkeley campus to evaluate.
© 2002 UC Regents. All rights reserved.