Extraction of oceanic features such as jets, fronts eddies etc, are of fundamental importance to Physical Oceanographers. Visualization and application of Computer Vision algorithms to achieve this goal has become popular and proven effective.

Computer vision (also referred to as "machine vision") refers to the automated extraction of information regarding the objects or scene in one or more images. The various problems faced in the field of computer vision are how to extract these features or information from a scene (Detection). How to find the boundary of the features (segmentation) in that scene and how to connect the same feature temporally (Tracking) for motion Analysis. The solution to all the mentioned problems is very domain dependent. This means that as we move from one working environment to another, our criteria for analysis changes.

This project of detecting Oceanic features from the MICOM dataset was started over three years ago. Since then we have also exported the data to the CAVE for visualization. We have also been working on AVHRR SST data.

Here you will find papers, presentations and student work on these topics conducted at the Vision Lab in the Computer and Information Science Department at UMD.