MEPS 312:297-309 (2006)  -  doi:10.3354/meps312297

ABSTRACT: In the future, if marine science is to achieve any progress in addressing biological diversity of ocean plankton, then it needs to sponsor development of new technology. One requirement is the development of high-resolution sensors for imaging field-collected and in situ specimens in a non-invasive manner. The rapid automatic categorisation of species must be accompanied by the creation of very large distributed databases in the form of high-resolution 3D rotatable images of species, which could become the standard reference source for automatic identification. These 3D images will serve as classification standards for field applications, and (in adjusted optical quality) as training templates for image analysis systems based on statistical and other pattern-matching processes. This paper sets out the basic argument for such developments and proposes a long-term solution to achieve these aims.

KEY WORDS: Natural object recognition · Object categorization · Zooplankton · Phytoplankton · Imaging · Taxonomy · Automatic identification · Image analysis