Robust particle outline extraction and its application to digital in-line holograms of marine organisms

Abstract

Digital holography offers a method of high-resolution imaging of microscopic particles and organisms in their natural environment. Automated image extraction and data processing are essential for rapid interrogation and analysis of the vast amounts of information contained in a typical hologram. In this work, we describe a robust-automated particle focusing approach, which we have developed to extract outlines of all particles contained within the sampling volume of each hologram constituting a "holovideo." The output data consists of ordered point-lists delineating polygons that match particle outlines and facilitate further processing such as extraction of focused images from the holograms themselves. The algorithm developed allows the reduction of, typically, a 2-GB holovideo to tens of megabytes, thereby greatly reducing analysis time by allowing rapid scanning of the contoured images without manual focusing. The algorithm has been demonstrated on synthetic and laboratory holograms and applied to holographic videos recorded in the North Sea. The algorithm output also lends itself to further automated analysis techniques like particle tracking or automated recognition.