Researchers at the University of Miami have developed a fully automated high-throughput specimen recognition software to assist the plankton imaging system, the In-Situ Ichthyoplankton Imaging System (ISIIS). The software uses novel machine vision and learning methods to achieve the best possible robustness to a high amount of data intricacies such as speckle noise, off focus effects, and shadowing, and identify and quantify a large number of specimens, whose shapes can be relatively similar to each other. The design of the software works on the principal that the most realistic situation for high-throughput organism recognition in plankton images is to assume imperfect segmentation of the existing organisms and use a robust classification scheme that can merge, split or exclude the detected regions of interests (ROI) during recognition. The system is applied successfully to automatically extract and classifies five target specimens, namely, copepods, larvaceans, chaetognaths, fish larvae, and Trichodesmium from the digital images.
|Number of pages||6|
|State||Published - Dec 1 2008|
ASJC Scopus subject areas
- Ocean Engineering