Digital image processing in the area of bioinformatics: Methods for segmentation and classification of biological features, through the example of selected diatoms
The geographical distribution of diatom species as well as the morphological characteristics of the same diatom species are the subject of current biological and paleoclimatological research. Computer assisted procedures can help to considerably accelerate the necessary analysis in this area of research. The presented paper covers two main points: First of all, the development of a practicable solution for a computer supported evaluation of mass screenings concerning light microscopic pictures of Fragilariopses kerguelensis. This technique identifies relevant valves, calculates their basic geometric data and matches them to a specific manifestation of shape. Secondly, the examination of algorithms for the computer assisted taxonomic classification of diatom valves. These are based on detailed structural analyses with regard to the problematic nature of three dimensional aspects and the application of different light microscopical procedures.