Estimation of surface wave parameters from image processing of video data

The National Weather Service in San Juan receives wave model output for the Puerto Rican regional coastal waters generated by NCEP's WaveWatch III model. These predictions are often unreliable mainly due to the fact that in-situ observations are not available for the coastal regions of Puerto Rico. These observations are required for calibration and validation of the wave model predictions. It is proposed to use video cameras to collect images of the nearshore zone of the ocean and to develop an algorithm for processing the video data. The data will be recorded in daylight by video cameras pointed at areas in the nearshore zone where the waves have not yet been affected by the reefs. After the video data have been received, a routine processes them in real time. To implement this, an observation system is required. Our design includes the following components: a Sanyo 1/2 inch CCD VGA (640/spl times/480) camera, an AXIS Camera Server, and a PC. The algorithm is based on several image processing methods and techniques, which allow for classification of objects with different grayscale intensity levels. The algorithm produces an output that includes estimates of wave parameters, such as period and direction in physical units.