Automatic, including in situ, image acquisition methods are explosively spreading in marine research, for investigating phyto- as well as zooplankton and other particles. Automated imaging has also become widespread in medical/histological microscopy, technically coming closest to the requirements of light microscopic diatom analyses. Building on these developments, we can now implement substantial parts of an automated diatom imaging workflow, resembling those drafted by the ADIAC project. To reflect the special characteristics of light microscopic images of diatom frustules (relatively simple closed outline shapes, low optical contrast), we developed an image segmentation and analysis tool which can be applied as a further, diatom specific component of high throughput diatom analysis workflows. Although a fully automated process enabling routine diatom counting is still far from being reality, we have successfully applied this workflow both in an intraspecific and in an intra-generic setting for morphometric analysis. Increasing the speed of analysis is thereby not our main motivation at the moment, much rather an improved precision in morphometric data capture and improved transparency of analysis results. In this talk I will give an overview of the current state of development of automated optical diatom analysis in our group, and highlight the potential as well as major challenges on examples from Southern Hemisphere species of the pennate diatom genus Fragilariopsis .