2018 Washington conference submission
Lossless JPEG2000 Conversion and Colour Profiles for Digital Preservation
Mel Mason - University of Oxford (United Kingdom), Andrew Hankinson - University of Oxford (United Kingdom)
Abstract: JPEG2000 is widely used as a delivery format for IIIF servers due to its intelligent decompression, and lossless compression makes it possible to use the same file as a long-term preservation format. However, digitisation tools and workflows still rely heavily on the TIFF format, necessitating a conversion from TIFF to JPEG 2000. For long-term preservation, we must ensure this conversion is lossless, both for the image data and for its associated metadata.
For our Digital Bodleian service we discovered a critical flaw in our conversion process that resulted in non-reversible conversions between TIFF and JPEG 2000 due to incorrect handling of colour profile information. To correct this, we re-examined our conversion process and set up a rigorous test suite to ensure our new method would not have similar problems. In this talk we will provide an overview of our new conversion process, designed to ensure fully lossless conversions between TIFF and JPEG 2000. Of particular note will be a discussion of how colour profile data should be handled to ensure images retain their original colour and are not inadvertently, and irreversibly, cast to an incorrect colour space.
We will also introduce a new open source Python library that converts a TIFF or JPEG to JPEG2000 using the Kakadu library, with an emphasis on digital preservation. This library includes automated checks that the conversion was completely reversible, on both the pixel and embedded metadata level.
Presentation type: 20 minute presentations (plus 5 mins questions)
Topics:
- IIIF and archival collections,
- IIIF-compatible software and experimentation
Keywords:
- JPEG2000,
- digital preservation,
- colour profile,
- Kakadu,
- lossless,
- Python,
- IIIF