Development of Automated Image Analysis Software for Suspended Marine Particle Classification.

摘要

The goal of this project is to develop a broadly-capable software package, which can automatically classify digital images of zooplankton. The software is being developed initially for classification of images from the SIPPER linescan imaging instrument, but has a wider application to alternate imaging systems where high quality digital images are available. Toward this end, the software is being developed to include both training and multi-stage classification portions. The input is digital images in standard computer format. A database-addressable classified particle list, including pertinent particle information and sorted images is the output. The project has application in rapid classification of microscopic marine particles, which have an effect on the instant optical properties and long-term variation in the local and global water column.