EurOCEAN 2000 - Vlaams Instituut voor de Zee

612
200 mm
stages with
optics and
CCD camera
1.9 m
1000 mm stage
45°
1.8 m
Plateholder
180mm
140x100x25
mirror
Laser / collimator
assembly
Figure 4: ‘HoloScan’ hologram replay machine (off-axis configuration)
• Physical Layout: Figure 4 shows a schematic of the replay system for in-line and off-axis
modes. The replay laser (Kimmon 180mW 442nm HeCd) is mounted above the collimating
optics (a reversed Galilean telescope producing a 100mm diameter beam flat to λ/5) along one
arm. A set of computer controlled stepper-stages (Ealing DPS system, 1000 mm stage with 10
µm steps and two 200 mm stages with 5 µm steps) carrying a videocamera (JAI CV-M300
CCD) is mounted along the other arm. The plateholder is near the vertex end of the stage arm
and it is possible to rapidly change the replay angle between the in-line (normal to plate) and
off-axis angles. The video output is captured by a ITI PC-Vision frame-grabber and then
processed to clean up the image and identify the true focal plane of each object within the 3dimensional
sample volume.
• Scanning and Data Extraction Procedure: The extraction of species distribution data is
separated into three steps. Firstly, the videocamera is scanned through the depth of the sample
volume (z) in a series of 0.1 mm steps. When the end is reached the camera is panned sideways
before returning to eventually scan the entire volume. For each step, the embedded image
processor receives the holographic images and processes them to clean up the noise, find
plankton micro-organism and locate the true focal plane of each object within the volume. The
system combines low (noise cleaning and image enhancement), middle (image segmentation,
object localisation and tracking) and high level (3-d information extraction) processing
modules that interact with a final high level module based on a neural net which allows species
classification.
System Trials and Performance
The camera is in the final stages of construction. A large water tank is currently being
fabricated which will encompass the front of the camera and allow completion of final
laboratory testing before dockside trials take place in August/September 2000.
A series of photographs of reconstructed holographic images of preserved marine plankton
may be seen in Figure 5. These were recorded in the laboratory in a Perspex tank using the
camera’s laser and optical assembly. The top row of "raw" images are taken from in-line
holograms (each image is 430 μm wide). The second row of "raw" images are taken from off-