| Popis: |
The Naval Oceanographic Office (NAVOCEANO) requires accurate estimates of seafloor roughness (bottom relief) and the density of seafloor clutter (mine-like echoes), typically derived from sidescan sonar imagery (SSI), to determine the bottom type of a geographic area for mine warfare. Determining clutter and roughness manually can be time-consuming and produce inconsistent results. Automated algorithms can derive clutter and roughness from SSI in a consistent and timely manner. Features such as pockmarks, sand ripples, and rocks on the seafloor are visible in SSI as bright spots ("brights") with adjacent shadows. The Naval Research Laboratory (NRL) developed a real-time clutter detection algorithm (transitioned to NAVOCEANO in 2001) that quickly and reliably identifies clutter in SSI and clusters the results into polygons. An object's height (estimated from the length of its shadow) is one measurement used to determine whether the object is mine-like. The authors theorized that height also could be used to automatically estimate seafloor roughness. NRL has developed a new automated roughness estimation algorithm, based on the clutter detection algorithm, to automatically derive seafloor roughness from SSI. In repeated trials, polygons generated by the new roughness algorithm correlated well (as high as 87%) with manually generated polygons for the same region. This article presents the NRL automated roughness algorithm (transitioned to NAVOCEANO in 2006), including test results and comparisons with manual methods. |
