Diatoms have become important in many different research fields. They serve to monitor the evolution of several ecological factors, they allow to judge water quality, they enable a sediment analysis in the search for gas and oil, they can be used by archaeologists to trace ceramic and clothing fragments, and they are even used in forensic investigations to establish the cause of death in the case of bodies found in water.
Diatom analysis always implies a sclanning of microscope slides and an identification based on morphological characteristics. This is a very tedious and complicated task, and in many cases a task performed by non-experts. For these reasons the automisation by computer processing would represent an enormous progress, freeing much effort for more important aspects like concentrating on e.g. ecological evolution.
This is the subject of the underlying proposal: we intend to automise the scanning of slide preparations in the search for diatoms, we will develop pattern recognition tools which allow to describe the complete diatom image, both the contour shape and the ornamentation of the valve surface, we will automise the identification by establishing representative diatom image databases, and we integrate the methods into an existing taxonomic database system.
This three-year project brings together specialists from diatom research in several applications, experts on pattern recognition and image processing, as well as experts developing taxonomic databases. It is meant to make diatom analysis an important and challenging application in pattern recognition, and to bring the European Community at the forefront of computerised diatom analysis.
Funding SchemeCSC - Cost-sharing contracts
9700 AV Groningen
EH3 5LR Edinburgh
NE1 7RU Newcastle Upon Tyne