Rapid reef assessment through subsurface underwater video
Abstract
We introduce the groundwork for the first automated rapid reef assessment system that uses a subsurface underwater reef video and a computer. Scale-invariant mean RGB color feature and Linear Discriminant Analysis (LDA) classifier were used for the system. Unlike in common rapid reef assessment protocols employed by marine scientists which take weeks to finish, data acquisition, through a subsurface video camera lowered from a boat, and processing can be done in one day. An overall recognition rate of 69.3 ± 18.2 % was achieved. The system's performance does not vary much with reef video depths of 1 to 8 meters, showing that color feature is suitable for the classification task. The application of rapid reef classification system is most promising for fast and frequent assessment and community studies of reef areas.