A Submersible Video Recorder for Marine Biological Surveys

We propose research to develop and demonstrate a deeply submersible video plankton recording system (ViPR) to perform image recording of plankton. The ViPR will apply artificial intelligence to 2-D or 3-D imaging using pattern recognition for a statistically reliable survey of plankton using two cameras, one for larger zooplankton and the other for microplankton. The ViPR computer will be PC compliant with a real-time operating system and a memory expandable to terabytes. It will also have processors to control the operation of the VPR, communications, and a new 3-D wavelet processing algorithm that will provide feature data to a Bayesian data fusion architecture that performs adaptive and trainable context based artificial intelligence to classify, identify and count plankton. The mono or stereo pattern recognition algorithms will process selected images from this sequence and combine each pair in such a way to classify, identify and count plankton populations with context data.