1995 National Preservation Office Conference
And There was Gold at
the End of the Rainbow ...
Developing an Architectural Image Database for Teaching, Learning and Research Activities at The University of Queensland.
Paper presented by Helen Darch, Library Research and Development Manager, University of Queensland Library at the 2nd National Preservation Office Conference: Multimedia Preservation - Capturing the Rainbow, in Brisbane, 28-30 November 1995.
- Abstract
- Background
- The Collection
- Options
- Methodology
- Descriptive Process
- Outcomes
- References
- Appendix A: Marc-based Descriptive Proforma
- Appendix B: Completed Marc-based Descriptive Proforma
Abstract
The University of Queensland Library, in collaboration with the University's Department of Architecture, has recently completed a pilot project to develop an architectural image database for teaching, learning and research activities at The University of Queensland.
Part of an extensive, privately owned collection of slides of Queensland historical buildings was used for the trial. The slides were deteriorating through age and handling. Several had disappeared while in use. The slides are a major teaching, learning and research tool and as many of the buildings represented no longer exist, the collection is also a significant record of Queensland's cultural heritage.
The problems addressed by this project are common to all libraries, museums and galleries, how to:
- preserve a unique collection of slides
- maximise access to pictorial collections.
The University of Queensland Library's solution was to preserve the slide collection by digitising the data. Access to, and use of the collection was maximised by:
- compressing the digitised data, storing it on hard disc on the Library's file server and providing University-wide access to it
- describing the pictorial data in detail and providing free text searching of that description.
This paper provides an overview of the trial.
Background
Universities are under increasing pressure to address the problems of how to preserve and maximise access to their pictorial (e.g. slides, photographs) and artefact collections. External pressure is being placed by government initiatives. Bodies such as the recently formed University Museums and Collections Review Committee are pressuring universities to 'find new methods of documenting, conserving and presenting their collections, collections which form a substantial part of the nation's cultural heritage'.(Campus Review 11/94).
The University of Queensland holds a wealth of unique, deteriorating pictorial collections. The breadth and depth of these are largely unknown. Increasingly departments are recognising the potential teaching, learning, and research value of their collections and are turning to the Library for advice on how to preserve and maximise access to these.
In late 1994 the Library began investigating imaging options. The literature was reviewed and imaging developments monitored by 'lurking' on Internet discussion lists such as Imagelib, DIGLIB and Syslibs. Several options were identified, none of which could be trialed until funding became available.
In November 1994, the Library and the Department of Architecture applied for, and gained, $A21,250 in funding from the University's 1995 Action Learning Program, to undertake a project that would preserve, and maximise access to and use of a sample collection of 1,000 architecture slides by digitising, describing and providing network access to that collection.
The Collection
One thousand 35mm slides owned by Michael Keniger, Head of Department of Architecture, and taken by him over a16 year period, were used for the project.
The slides focus on Queensland historic buildings ranging across domestic, public, mining and agricultural buildings, many of which have now been demolished. As a number of the buildings have not been recorded elsewhere, the collection is significant as a record of the nation's cultural heritage, as well as being a heavily used teaching, learning and research resource.
Potential copyright problems were minimised as the slides were privately owned by one of the project partners.
Options
Two approaches to digitising pictorial materials were identified, the 'do it yourself ' and 'pay someone else to do it' options. Cost, lack of in-house and the findings of other institutions that it is cheaper to contract for digital conversion services than to 'do it yourself' (Conway, 1994), influenced our decision to take the 'pay someone else to do it' option.
After preliminary investigations the SearchTech system was chosen. SearchTech digitises the material, provides a mixed text and image database, user-friendly interface and rapid response time without the need for major hardware expenditure, all at a relatively low cost. It is in use in a number of Australian sites including the National Library of Australia, State Libraries of New South Wales, Victoria, and the Northern Territory, and the Australian War Memorial.
Kodak's Photo-CD system was not seriously considered due to its high cost and concerns on the Imagelib Internet discussion list re Kodak Photo-CD's quality control.
The SearchTech system combines broadcast quality imaging (736 x 560 pixels) in 16.7 million colours, JPEG compression and Windows compatible software, to offer a fast, economic, text-based image retrieval system. A broadcast quality video camera is connected to a frame-grabbing board and the digitised images are compressed using the JPEG standard compression technique.
The JPEG standard 'allows image file sizes of 1Mb to be compressed to approximately 45K, transmitted across 10Mbits/second Ethernet LAN and decompressed within three seconds on a standard 486 PC '(Walker, 1995). The JPEG decompression routine used is Express Compression Laboratories', one of the fastest on the market.
The JPEG compression rate used is 25:1. The high compression rate used allows up to 20,000 images to be stored per gigabyte. We have some concerns that the compression rate may be too high. Other studies suggest that it is acceptable but should not be exceeded for 'maximum trade off between image quality and file size' (Mason,1994).
SearchTech uses QCOM's Corporate Retriever as its search engine, with Perfect Pictures providing the Windows-based user-friendly front end to both Corporate Retriever and the Express Compression Laboratories JPEG decompression routine.
No specific hardware is required. Equipment purchased for the project consisted of 64mb memory for the Library's file server, a 486 DX2/66 personal computer, a 1Mb video card (2Mb would have been better), and a Hewlett Packard 1600cm desk jet colour printer.
The system is easy to use and most users require no training. One enters a search term and the system responds with a count of the number of matching images. The user can then browse through those records, simultaneously viewing an image and description of that image. Full screen size display of the image only is possible. Images can be printed on a colour printer or downloaded onto a floppy disc and used with other software packages such as Adobe Photoshop which allows the user to manipulate the image.
While SearchTech's image quality is not as high as some other systems, it represents a fast, cheap solution at only 95 cents per image and 15 cents per descriptive link.
Methodology
The project consisted of a number of stages.
Stage one involved selecting and sending two batches of 500 slides to SearchTech for conversion to digital bit mapped image and JPEG compression. Within a four week period the slides and their digital surrogates were returned as a series of discs which were then copied to hard disc on the Library's file server.
Stage two, the descriptive process, was the most problematic and time consuming. A research assistant (an architect) was employed for five months to work in collaboration with Library staff to develop a standardised description for each digital surrogate.
Stage three involved:
- loading the SearchTech system software onto the Library file server
- entering the digital surrogate descriptions on a word processing package (we used Microsoft Works which necessitated converting the data into ASCII format)
- setting up the Corporate Retriever software (no mean feat as no documentation was provided)
- using Corporate Retriever to index the text files.
Indexed files were then used by the Perfect Pictures software which acts as a front end to both the Corporate Retriever search engine and the image display software, and provides the Windows-based user friendly front end.
Stage four was a two-step process.
A two hour pre-trial test was held using a group of Architecture students with varying levels of technical skill. The test proved a useful means of validating image descriptions, checking the user-friendliness of both the system and the documentation we developed on how to use it, testing the capacity of the system when networked and under load.
A number of problems were identified and most resolved prior to the official trial currently underway in the University's Architecture/Music Library. For the duration of the trial an evaluation form has been available next to each access machine. It seeks information on what the user was looking for and whether it was found, as well as comments on the usefulness of the system and the quality of image resolution.
Stage five, December 1995, will involve enhancing the descriptions of the digital surrogates based on feedback obtained from the trial. Network access will then be made available to the University's staff and students via the Library network.
There will be a stage six as funding has just been received from the University's 1996 Action Learning Program to:
- convert the database and make it available via the Library's World Wide Web server
- expand the amount of material on the database
- include three dimensional material on the database.
Descriptive Process
The first descriptive challenge was naming the project. DIGILIB (i.e. DIGItal LIBrary) was selected as an umbrella title suggestive of a series of digital text/image databases. Hence the project became known as DIGILIB - Queensland Country Towns database.
Images are only of use if one can find them. A primary aim of the project was to enrich or 'value add' to the slides by providing the best possible image descriptions. This aim provided multiple challenges for in order to translate visual information into verbal information one must describe not only the image's primary object, its relationship with other objects, but also the conditions of the image's creation.
Descriptions were developed under the guidance of the project's reference group. The reference group consisted of three experts, one on Queensland architecture; one on Queensland history; and another on the history of costume. As a preliminary descriptive exercise, the reference group was shown a selection of slides and asked to describe their content.
Each saw the same slide in quite different ways, a timely reminder of the impossibility of predicting all required search terms. This exercise confirmed Enser's findings that images mean different things to different people, that they will even mean 'different things to the same person at a different time' (Enser, 1995) and validates Cawkell's theory that image description involving multiple experts with different perceptions greatly enhances the quality of the description and maximises its usefulness (Cawkell, 1992).
A Marc-based description proforma was developed to provide an interpretive structure that would guide the descriptive process. Appendix A shows the proforma and Appendix B, a completed example.
The descriptive process represented the major cost of the project. Our findings were comparable with Keefe's, that it takes an average of 40 minutes to index an architecture slide (Keefe, 1990).
Descriptions were based on the Visual Resources Association's Data Standards Committee Working Draft and the Library of Congress' Thesaurus for Graphical Material. Specific architectural terms were taken from the Art and Architecture Thesaurus and Richard Apperly's A Pictorial Guide to Identifying Australian Architecture: styles and terms from 1788 to the present (North Ryde, N.S.W.: Angus and Robertson, 1989). Australian historical terms were based on the Australian Heritage Commission's HERA Thesaurus.
A list of inhouse descriptors was developed to cover local conditions and in an endeavour to bring a measure of control to the descriptors used. The thesaurus consists of two parts, a straight alphabetical listing of all descriptors used, and a list of descriptors sorted into subject groupings.
Australian spelling was used and efforts have been made to provide a 'see' reference to link Australian and American spelling (e.g.Moldings USE Mouldings).
Internationally recognised standards were followed where possible to ensure compatibility with the Library's Library of Congress based catalogue, maximise continued access to the material and minimise maintenance problems.
Appendix A shows the descriptive elements provided for each digital surrogate.
The SearchTech system creates a brief database 'link' for each slide scanned. One can then either pay SearchTech to enter the text or do it oneself. We chose the latter, cheaper option and lived to regret the decision due to the lack of information on how to 'do it yourself'.
Once the SearchTech software has been loaded onto the system, authorised library staff can expand the 'link' into a bibliographic record that contains a maximum of 50 named variable length fields. Descriptions can be up to five A4 pages in length and all words (other than library defined 'stop' words) are indexed, thus searchable. Free text searching is available and boolean operators can be used.
Outcomes
The project had multiple outcomes. It produced:
- a potentially marketable integrated text and image database that has provided University-wide, concurrent, multi-user access to a previously restricted teaching, learning and research resource
- a standards-based proforma that can be used as a prototype for anyone describing images
- a standardised thesaurus for the description of images of architecture
- an evaluation form that can be used for the evaluation of image databases.
While dramatically increasing our knowledge of imaging issues, the project raised more questions than it answered. Will easy access to digital surrogates increase or decrease demand for the originals? How will we institutionalise migration and storage strategies? How can we preserve copyright and intellectual property in a digital age where it is easy to manipulate original data, change its context, integrity and authenticity?
As with the rainbow, we struggle over each hill only to find yet another on our trek to that elusive pot of gold.
References
Conway P 1994, 'Digitising', Library Journal, Feb 1 pp.42- 45
Cawkell A.E. 1992, 'Selected aspects of image processing and management: review and future prospects', Journal of Information Science, v.18 p.182).
Enser P.G. 1995, 'Pictorial information retrieval', Journal of Documentation, v.53 no.3 June p.145)
Keefe J 1990, 'The image as a document: descriptive programs at Rensslaer', Library Trends, v.38 no.4 1990 pp.659-681
Mason P R 1994, 'Imaging system components and standards' in Elkington N.E. ed. Digital Imaging Technology for Preservation, Proceedings from an RLG symposium, Cornell Uni, Ithica, N.Y., 1994
Walker, I 1995, 'Pictorial Management Systems' Information Online and On Disc, Proceedings of a conference, Sydney, Australia, 1995 p. 312
Appendix A: Marc-based Descriptive Proforma
Field name Marc Field content
Code
FILE (001) File number
Slide record number
DATE (008) Date of creation
Date or dates of construction
DESIGN (100) Main creator of work
Architect - or if builder - explain in notes section
TITLE (245) Main title of work
Title of object and Name of town
PUBLICATION (260) Creator of image
Photographer and date slide taken
DESCRIPTION (300) Technical description
Original medium, scanning compression, resolution,
where scanned and date
SERIES (440) Collection
Department - Collection Title
NOTES (500) General notes - expand or add info in any field
which doesn't fit into other categories
May include re-use of building, orientation,
historical information, references to other images
of object
COPYRIGHT (500) Holder of copyright licence
Name and address
COPY.RESTRICT (500) Parameters for copyright restrictions
OBJECT (650) Subject
Generic title of objects in photograph - refer
thesaurus
USE (650) Subject
Specific title of objects in photograph - refer
thesaurus
KEYWORDS (650) Subject -
Features, Condition, Structure, Materials,
Interior, and context - refer thesaurus
PERIOD/STYLE (650) Subject
Historical period and style - refer thesaurus
LOCATION (651) Location of image
Street, Town, Latitude, Longitude, Country
ADD CREATOR (700) Additional creator
Added building owner, builder etc - explain in
notes section
VARIANT TITLE (745) Additional titles
Popular or other names - explain in notes section
Appendix B: Completed Marc-based descriptive proforma
Field name Field content FILE 0000435
DATE c.1927
DESIGN Elina Mottram
TITLE Masonic Temple Longreach
PUBL Michael Keniger 1979
DESCRIP 35mm original - JPEG compression - 736 x 560 pixels -
SearchTech June 1995
SERIES Architecture - Queensland Country Towns
NOTES Elina Mottram was the first female architect to open her
own practice (1924) Also FILE 668
COPYRGT Michael Keniger - Department of Architecture - University
of Queensland
COPYRES copy for educational purposes only
OBJECT Recreation and sports facilities Streetscapes
USE Clubhouses Masonic halls Urban streetscapes
KEYWORD (Features)Balconies Cornices Doors Entrances Parapets
Pilasters Verandahs (Condition)(Structure)Brackets
Masonry construction (Materials)Glass Rendered brick
(Interior)(Context)Trees
PERIOD Federation Academic classical
LOCATE corner Galah Street and Duck Street Longreach 23.28S
144.14E Australia
DESIGN2
TITLE2
Last updated 27 February 1998
| About This Site | Copyright | Privacy | Accessibility | Site Map | Site Search | Content A-Z | Contact Us |