A t Auto mate d Pre se rvatio n: t d P ti T he Case o f - PowerPoint PPT Presentation

A t Auto mate d Pre se rvatio n: t d P ti T he Case o f Digital Raw Pho to graphs Stephan Bauer, Christoph Becker ICADL 2011 Beijing j g

Joseph Nicéphore Niépce

stallio (flickr)

Why RAW digital negative digital negative most authentic version of the image no lossy compression as in jpeg new tools may create better interpretations y p not an image: uninterpolated sensor data

De ve lo ping a raw file Demosaiquing White balance adjustment White balance adjustment Colorimetric interpretation Colorimetric interpretation Tone mapping i Image enhancements

interpretation

r onverter e DNG C Adob Adobe dcraw Apple pp

validate migrate

E valuatio n frame wo rk numerous proprietary raw formats – high risk numerous proprietary raw formats high risk normalization to standardized format desirable How to evaluate and validate? preservation planning preservation planning systematic evaluation method and tool controlled experimentation controlled experimentation automated measurements

migrate comparison comparison comparison comparison

Auto mate d Me asure me nts traditional quality assurance methods traditional quality assurance methods are error based common tools not reliable common tools not reliable requirements perception based respecting ICC-profiles meaningful for color images

equal MSE https: / / ece.uwaterloo.ca/ ~ z70wang/ research/ ssim/ # MAD

Signific ant Pro pe rtie s - Co nte nt SSIM relative AE SSIM Hue relative MSE SSIM Saturation

Signific ant Pro pe rtie s - Co nte xt IPTC Exif (exposure) Dublin Core Exif (technical) Private Tags Exif (location) Metadata XMP Exif (generated) (g )

Co ntributio n image comparison metrics implemented image comparison metrics implemented using Java Advanced Imaging API metadata metadata verification using ExifTool e ification sing E ifTool comparing Adobe DNGConverter and digiKam p g g migrate and validate a representative selection

DNG by digikam incorrect color matrix generated raw data are identical raw data are identical Re sults – CRW RAW

RAW RAW DNG by digiKam Re sults – CRW

ADC DNG DNG by two embedded color matrices Re sults – CR2 RAW RAW

DNG by ADC RAW Re sults – CR2

Co nc lusio n demonstrated fully automated QA using state of the art perceptual level method i t t f th t t l l l th d well suited to falsify: find bad conversions yet not suited to fully verify conversions SSIM allows meaningful measurements

Outlo o k implement workflows using taverna implement workflows using taverna run large scale tests correlation of SSIM and manual evaluation use more tools for QA Q expand approach to different types of content

? Automated Preservation: The Case of Digital Raw Photographs Stephan Bauer, Christoph Becker, ICADL 2011 www.ifs.tuwien.ac .at/ ~bec ker