In SAR imagery application for ship detection, ship backscattering along with anomalies in the background (figure 1) has been calculated. Ships are generally constructed by large flat metal sheets and are radar bright and effortlessly detectable in SAR imagery (figure 2).
Our system has a hierarchical design to extract ship features and recognizing ships from SAR imagery. Our Ship-detection-system segments, detects, classifies and declares categories of ships using TerraSAR-X imagery robustly. The system performs segmentation of ships from the ocean clutter and generates silhouette by ship centreline-detection algorithm. Ship category extraction is based on the radar scatterers distribution in ship’s nine sections along the ship’s range profile. A three-layer neural network is trained on simulated scatterers distributions and supervised by a rule-based expert system at the background. Ship category is subsequently estimated by a Bayes classifier based on the ship length. It is noticed that segmentation and centreline detection are sensitive to ship aspect angle because of the diverse radar scattering signature.
Our experience shows that X-band imagery in HH polarization is preferred for detecting ships as ship-sea contrast is higher for HH polarizations. Conversely VV is preferred for wake detection. As lower backscatter at HH decreases rapidly with increasing incidence angle, ship wakes are rarely identified in HH polarized imagery (figure 3). However localised wakes will be problem for ship orientation detection (figure 4). Larger incidence angles give better ship discrimination in high wind conditions. SAR image intensity depends on the direction of the wind relative to the SAR look direction.
TerraSAR-X imagery for Ship-Detecting System
The capabilities of TerraSAR-X Strip (GSD 6m) and SPOTLIGHT SAR (GSD-1m) data are investigated. TSX can acquire HV polarization only for StripMap and the resolution is not enough to extract subtle features. However, category of ship could be recognized (Level-I). While processing SPOTLIGHT mode imagery, it was observed that for ship identification HH and for wakes VV polarisation produced good results. Our Ship-Detection algorithm considered the properties of TSX imagery such as backscattering values, resolution, incidence angle, polarization, Radar frequency. Other relevant factors considered are, pixel spacing, number of looks, and swath width.
In our system (figure 5), the following procedure and conditions are customized.
Summary
The ship-detection system shows that several features of ships can be obtained from TSX- SAR imagery. Both line and merchant category ships can be identified and classified from their size and RCS values. The system is very simple and robust (figure 8). Usually ship scattering study should be done with simulated scattering maps. Since the first author did not have access or facility to perform large research that involves several data collection, a representation-database used is based on literature. We developed a fundamental and robust TSX-SAR imagery modelling for Ship detection. It is recommended to exploit original classification or generally accepted classification for a particular country.
Acknowledgements
The authors thank PASCO-Japan and Infoterra for data and for encouraging this research work.
Biography of Authors
Babu Madhavan obtained PhD (1995) in Remote Sensing (SAR) from IIT-Bombay. Later, completed two post-doc research projects at Keio University, Japan and joined PASCO in 2002 and served as CEO for its office in India (2005 to 2010). With 24 years work experience in image processing and computer vision research, Babu published more than 85 papers. Recently Babu won ‘Geospatial World Innovation Award’ for 3D modelling technology development in The Netherlands for his research at Softopia Japan.
Dr. Tadashi Sasakawa Vice-President of PASCO, obtained PhD.Eng.(2005) from Hokkaido University. He joined PASCO in 1982 and remained engaged in dissemination of GIS. In 2000, he was instrumental in the introduction of aircraft sensors ADS40, UCD, LIDAR, and automation of digital photogrammetry. In 2005, he established Satellite Business Division for commercialization and applications of TerraSAR-X imageries.