National Library of Australia Staff Papers, 1999

Font Size:  Small  Medium  Large

Rare Maps Digitisation Project

Paper presented by Maura O’Connor, Map Curator, National Library of Australia at the Australian Map Circle 27th Annual Conference, Brisbane, 5-8 April 1999

Background

Four rare or pre-1900 map collections relating to the development of Australia’s cartographic knowledge have been acquired and maintained by the National Library of Australia since 1901. Regular acquisitions continue to be made to one of the collections, with the total of all four collections now numbering between 7-8,000 titles. Three of the collections were established by the founding collectors of the Library’s collections : Petherick (Rare Map), Ferguson and Rex Nan Kivell. A fourth collection, the Tooley rare map collection, was purchased in the early 1970s.

Full bibliographic control has been established for both the Ferguson and Rex Nan Kivell collections. Partial control has been achieved with the remaining two collections.

Access to the collections is limited to personal visits or the ordering of photographic copies, sight unseen, as the material is not permitted out on Inter-Library Loan, due to fragility and rarity. In recent years, a growing demand for access to this material from researchers and technological advances such as digitisation and the Internet suggests now is an appropriate time to make the collection more accessible by providing an alternative copy of this rare material.

Improved access to the Library’s rare map collection has been the impetus for conducting a digitisation project of the material. The success of other digitisation projects in the Library, notably the Images 1, together with the Internet and the Library’s automated catalogue, Kinetica also provides options for improving access to this collection if the digitisation issues relating to early maps can be successfully addressed. Direct access to a map database with links between the catalogue record and the image, such as in Images 1, may be a possibility while an integrated approach, using all three technologies, provides another possible option.

Preservation concerns which have been the impetus for other rare map digitisation

Projects, notably the Utzon project at the Mitchell Library, are currently less an issue than the access consideration. However, in the future, as the technology improves, there may be potential for long-term surrogates in the digital image which will provide a useful alternative if a natural or human disaster should destroy the originals.

In March 1998 following on from the success of the Library’s Images 1 project to digitise(scan) part of the Library’s Pictorial collection, a proposal was put forward to digitise(scan) the Rex Nan Kivell rare map collection. Not only is it under full bibliographic control but it would also complement the Pictorial component of that collection already digitised under the Images 1 project.

The Trial Project

Objective

Technical reports from similar projects being conducted overseas, notably the Library of Congress, University of Georgia and the British Columbia Archives and Record Service, were examined in order to identify known problems and to what extent they could be avoided within the limits of the technology available.

A project was subsequently designed as a trial or experiment on a number of techniques available to determine:

  1. The most effective means of creating high quality digital scans of rare maps.
  2. To determine the appropriate resolution levels required to provide satisfactory viewing levels.
  3. To link digital scans to bibliographic records and mount a test database on the Internet.

Methodology

Twenty-five maps from the Rex Nan Kivell rare map collection were selected representing apsects of the following criteria:

  • range of sizes available
  • black and white, or colour
  • physical condition, including any particular problem such as dissected and
  • mounted, bound into books, mounted on board, tape or other repairs, tears.

Following an assessment made of rare map digitisation projects undertaken elsewhere, notably Library of Congress, Mitchell Library (Utzon plans Preservation Project), and the Library’s Images 1 Project, a proposal was put forward to conduct a trial comparing scanning of photographic images with direct digital scans.

MrSID (Multi-resolution Seamless Image Database), produced by LizardTech, was chosen as the compression solution to trial during the project. Using wavelet compression it has been used by the Library of Congress in its own map digitisation project. MrSID images can be made available, using a MrSID server over the Web, or downloaded for local viewing using a MrSID viewer available free from LizardTech website.

Three options were explored:

Option 1

All maps were photographed on 4” x 5” transparencies and scanned using professional PhotoCD at Base64 by the Library’s contractor for routine digitisation programs (Images 1). Once Professional PhotoCD was received, the highest resolution available for each map was used to create a TIFF image using Photoshop 4.0 Pro. Full size of 72 megabytes with a pixel size of 4090 x 6144 was achieved. MrSID was then applied to compress the images.

Option 2

(i) A 12mb file capacity digital camera was hired for one month and used to scan from the original maps. For the larger maps, scans in sections were also produced. The camera created TIFF images of 2048 x 3072 pixels when written to PhotoCD, creating files of approximately 18 megabytes each. MrSID was then applied to compress the images.

(ii) A 129mb file capacity digital camera was hired for one day following an assessment of the results from the previous camera, which were insufficient, and a subset of three maps was digitally captured. TIFF images of 6000 x 1520 pixels were created with files of approximately 129 Mb each (RGB export file). MrSID was then applied to compress the images.

Option 3

Inhouse scanning from the 4” x 5” transparencies were also created and written to photoCD. TIFF images were created, of approximately 18 mb each and 2048 x 3072 pixels. MrSID was then applied to compress the images.

Three inhouse scans at various levels of resolution, ie 500dpi, 800dpi and 1000dpi were also produced of three of the images.

Results

Option 1

Scans produced from 4” x 5” transparencies using the Library’s routine digitisation procedures did not consistently produce images of a high enough resolution to provide sufficient detail for use by a researcher. This was the case for both uncompressed, TIFF images and for compressed MrSID images. The resulting images tended to be dark, dull or faint and exhibited low levels of contrast and brightness.

If transparencies were to be considered as an option further analysis would be necessary to determine the maximum map size that could be successfully digitized this way. In addition, further analysis would be required to determine whether some maps within this size range would still not digitize satisfactorily. This may be due to map detail and/or colour.

The administration and clerical resources need to prepare maps for processing under this option were considerable and time consuming.

Option 2

(iii) For maps digitized using the 12mb digital camera, the quality and resolution was inadequate. Even when the optimum pixel size was achieved (based on map size), the clarity and robustness of the image was unsatisfactory. Text was unreadable with the image and colour dark and blurred.

(iv) The 129mb digital camera with resolution level capacity of 300dpi, produced the optimum quality – clarity, robustness, contrast and brightness were all good by comparison with the images produced by scanning transparencies.

No perceptible loss of detail or data quality was observed with the display of the high resolution images (300dpi) in the MrSID viewer. This was ascertained by comparing the original photoCD and/or TIFF image (viewed in PhotoShop) with the compressed images.

Option 3

Scans from inhouse transparencies did not convert to satisfactory TIFF images.

A series of trials using higher resolution levels ie500dpi and higher from transparency scans, either 4” x 5” or 10” x 8” did not produce satisfactory results. It would appear that transparencies do not produce a suitable resolution level for graphic images.

Discussion

Most effective means of creating high quality digital scans of rare maps

Results of scans from the transparency images, and the direct electronic image, clearly indicated a more effective result from the latter. A far more legible image was achieved, not afected by the size of the original maps, although with very large items, several, non-overlapping images were required (up to four sections) to ensure legibility was maintained.

The choice of resolution capacity in the electronic camera proved to be the next issue once the choice of scanning option was resolved. The 12mb digital camera did not provide the required quality and resolution levels to ensure legible viewing either before or after the compression process. In contrast, the 129mb digital camera, with its higher resolution facility provided a more satisfactory result even though tried on a very small subset of three maps from the original selection.

Consequently, of the four options available, use of the 129mb digital camera for direct scanning of the maps is the preferred option together with the appropriate resolution level (300dpi) and supporting imaging software to give the computers capacity to maintain colour quality and speed of image management.

A consideration under Option 1 of using the transparency process for maps up to a certain size would require considerable administration and high staff resourcing to sort out and select maps prior to filming. It is not considered a viable option.

Determine the appropriate resolution levels required to provide satisfactory viewing levels.

Considerable testing was undertaken of the transparency scans and the two sets of electronic scanning cameras to find the optimum resolution level required.

From the results obtained, it would appear that a resolution level of 300dpi or above is required to provide the most effective image for viewing. Lower resolution levels produced by the 12mb camera were not suitable. Increasing the resolution levels on the transparency scans did not improve the quality of the image inferring that transparency scans of graphic images do not produce a suitable result.

Link digital scans to bibliographic records and mount a test site on the Internet

Problems were encountered with the MrSID Server software supplied by Lizard Tech in that it was not compatible with the Library’s AIX/UNIX platform. While this problem has now been resolved it was not possible to do so in the trimeframe of the trail project. Consequently it was not possible to test run the images over the Internet and link to the bibliographic records. The next stage of the project will test the feasibility of this process.

Issues arising out of the Trial Project

The Trial project demonstrated that digital scanning of early maps is feasible with a number of options available. The most viable and appropriate option is that of direct electronic scanning, ie with a digital camera, rather than transparency scans. From the Trial, the cameras produced the best result in terms of clarity, resolution and quality (colour).

A minimum resolution level of 300dpi was determined as the most appropriate level of resolution. Minimum technical standards need to be developed and confirmed to ensure the quality of the digital image is maintained irrespective of the size and complexity of the original item.

Insufficient time was available to link the bibliographic records to the images although this was not seen as a particular problem or concern.

Further Development

A two-tiered Pilot project is proposed for the 1999/2000 financial year which will build on the results of the Trial project carried out in 1997-8.

The Pilot project will determine:

  • Confirmation of the need for 300dpi resolution for rare maps.
  • Develop standards for technical requirements for all images, including colour
  • modelling and reproduction accuracy
  • Develop technical solutions for maps requiring sectional scans and scan a
  • sample of 25 maps from the Rex Nan Kivell rare map collection
  • Mount the sample on a demonstration web site for public comment with links
  • to each image’s bibliographic record.
  • Develop a cost profile for the project and clear procedures for routine rare map
  • digitisation.

Once the Pilot project has been successfully completed it is envisaged that a two year project will then be developed to complete the digitisation of the Rex Nan Kivell rare map collection.

The developmental work currently being undertaken with respect to the Rex Nan Kivell rare map collection will pave the way for routine digitisation of the Library’s rare map collections in the future. The development of standards and procedures will ensure that the processes, which are complex and experimental at present, will become routine and greatly improve the access to a collection which for so long has been limited.

References

Allen, David Yehling 1998, ‘Creating and distributing high resolution cartographic images’, RLG DigiNews, 15 August, vol. 2, no. 4 . At : http://www.rlg.org/preserv/diginews/diginews 2-4.html

British Colombia Archives. Cartographic records pilot project. At : http://www.bcarchives.gov.bc.ca/cartogr/general/maps.htm

Columbia University. Oversized color images : addressing issues of preservation and access. At : http://www.columbia.edu/dlc/nysmb/reports/phase 1.html

Henneberger, Bob 1994, ‘Preservation and access of rare maps at the University of Georgia libraries’. Microform Review, vol. 24, no. 4, pp. 169-171

Library of Congress 1996, ‘Cartography : Industry support moves Library to state-of-the-art digitisation’. National Digital Library Program, No. 11, September.