Plankton Detection and Tracking Using Frequency Filtering

Sophisticated optical hardware allows performing computationally heavy image processing with the speed of light, before the actual frame is captured by a camera sensor. Filtering in the frequency plane of the optical system can be used for fast pattern detection, which constitutes one of the major problems in plankton image processing. The aim of the research presented in this paper was to design a set of spatial filters allowing for particles detection and tracking in video streams. The method enables in a limited scope to distinguish the species of the tracked organisms. This initial recognition is based on the size differences.