Sponsors: Office of Naval Research
The OmniCam is an omnidirectional, high resolution, high speed, underwater imaging system designed to record the radiometric light field in the open ocean. These data are important in understanding the temporal and spatial fluctuations of the ambient light field. In addition, these data can be used post-facto to study the response of light sensitive animals (such as squid) in a controlled laboratory setting by replaying data recorded in situ. In addition to its use in light field measurement, the Omnicam system has applications in wide field of view surveys of coastal ocean environments such as reefs and kelp forests.
The OmniCam system includes six, 2-Megapixel HD CCD cameras made by Allied that are arranged to form six sides of a cube. Each camera is supported by a VIA EPIA-P720 Pico-ITX computer and solid state drive. This allows each camera to record data in parallel allowing for simultaneous acquisition of uncompressed HD video at up to 20 fps on each camera. Each camera has a 2.7 mm Fujinon fisheye lens (FujiFim) to provide overlapping views of the light field.
The system is housed in a 44cm diameter Vitrovex glass sphere (Nautilus Marine GmbH) and stands 2ft tall. This glass housing contains 10 protruding connectors; the top 6 provide high-speed USB links to readout the collected data, while the bottom 4 provide charging capability, Ethernet connectivity, and 2 serial communication ports and a conduit for DC power.
Together with the titanium frame and tracking systems, OmniCam weighs 124.5lbs (56.47kg) in air and is ballasted to be slightly negative in water; it was configured for autonomous deployment to depths of 6000 meters.
OmniCam was designed to capture and store up to 180 million pixels per second of 14-bit uncompressed data. The system is capable of writing image data to disk at a 60 MB/s sustained rate for each camera, giving a total throughput of 240 MB/s. This allows uncompressed 1920 x 1080 images with a 14-bit depth from each camera to be collected at an uninterrupted rate of 15 fps (alternatively, 3×3 binned images at 100 fps).
The entire system is controlled by a set of in-house developed programs that operate in parallel to synchronize all of the cameras, store the collected data from each camera and all auxiliary instruments, and carry out programmed deployment scenarios. The camera-streaming program was developed in C++ using Allied Vision Technologies SDK.
- Principal Investigator: Jules S. Jaffe
- Principal Engineer: Fernando Simonet
- Software Engineer: Ben Laxton
- Optical Consultant: Paul L. D. Roberts
- Graduate Student: Justin Haag
- Lab Assistant: Wing Cheuk
Photos, Videos, Data: