Automated Texture Recognition of Quartz Sand Grains for Forensic Applications

Authors


  • Funding support for the work undertaken during this study was provided by AWE Plc.

Additional information and reprint requests:
Ruth M. Morgan, D.Phil.
University College London Security and Crime Science
35 Tavistock Square
London
WC1H 9EZ
U.K.
E-mail: ruth.morgan@ucl.ac.uk

Abstract

Abstract:  Quartz sand surface texture analysis has been automated for the first time for forensic application. The derived Basic Image Features (BIFs) provide computer-generated texture recognition from preexisting data sets. The technique was applied to two distinct classification problems; first, the ability of the system to discriminate between (quartz) sand grains with upturned plate features (indicative of eolian, global sand sea environments) and grains that do not exhibit these features. A success rate of grain classification of 98.8% was achieved. Second, to test the ability of the computer recognition system to identify specific energy levels of formation of the upturned plate surface texture features. Such recognition ability has to date been beyond manual geological interpretation. The discrimination performance was enhanced to an exact classification success rate of 81%. The enhanced potential for routine forensic investigation of the provenance of common quartz sand is indicated.

Ancillary