Data Extraction from Underwater Holograms of Marine Organisms

A digital holographic camera (eHoloCam) has been developed by us for in situ underwater studies of the distribution and dynamics of plankton and other marine organisms and particles. The eHoloCam uses a frequency-doubled Nd-YAG pulsed laser to freeze-frame rapidly moving particles and a CMOS imaging sensor for high-resolution image capture. Digital holograms are recorded at rates from 5 Hz to 25 Hz as holographic videos over a period of several hours. Data is stored in the camera on an embedded computer. eHoloCam is capable of recording all organisms and particles located in a water volume of 36.8 cm³ in a single hologram frame. The recorded holographic videos may subsequently be recreated numerically, using a variety of reconstruction algorithms, at a desired image plane. The vast amounts of data stored in holographic videos presents a major challenge for the automation of image extraction, identification of species and hence analysis of the holograms. In this paper, we describe some of the algorithms we use to optimise hologram reconstruction and, subsequently, image quality. We discuss the problems of automating the data extraction, the implementation of auto-focus methods and the definition of regions of interest in the eholovideo reconstruction procedures. In covering each of these and their amalgamation into a single algorithm, we will describe a complete image auto-extraction algorithm capable of scanning eholovideos and generating a directory of extracted, focussed bitmap-images. We go on to discuss possible analysis techniques that could be applied to resulting images in order to aid the extraction of particle information such as particles-per-volume and particle size ranges, and finally discuss the future implementation of particle auto-recognition to perform species-specific analysis of eholovideos.