Ship detection in the Brazilian coast using TerraSAR-X SAR images

Extensive Brazilian coast and a growing up maritime vessel traffic lead to research better ancillary methods to control ship´s flow. Orbital SAR imagery is an important tool, especially due to its ability to work day and night and clouds coverage doesn´t interfere often. At present study, some well-known ship detection concepts are investigated, applying them to TerraSAR-X (TSX) ScanSAR images (16 m resolution), in VV and HH polarization. Statistical parameters are estimated and trough Kolmogorov-Smirnov test K-distribution fit to TSX images is checked. Algorithm detection based on Constant False Alarm Rate (CFAR) concept is applied and its performance is verified. Incidence angle, CFAR´s windows size and Probability of False Alarm (PFA) influence are further analysed. Finaly, we discuss some operational issues about near-real-time image delivery and the delay impact on a surveillance system or on an urgency request.