Ocean monitoring and surveillance has gained great interest and importance in recent years owing to the increasing maritime concerns where examples include piracy and illegal fishing. Ship detection using satellite synthetic aperture radar (SAR) has become a widespread choice, due to its different advantages which include the all-times/all-weather operational capability, high resolution and wide coverage. The first SAR generated product is a Single Look Complex (SLC) product; a two-dimensional array of complex valued pixels containing amplitude and phase information. It is common practice in ocean monitoring and ship detection that SLC products are multilooked where only amplitude information is retained and used for ship detection. However, valuable information contained in the complex imagery may be lost when only using the generated amplitude data for ship detection. We argue that information in the complex data is significant and highly useful for ship detection applications. This lecture proposes to shed light on this issue in satellite SAR ship detection. We first review the process of SAR image formation and ship detection in SAR imagery, second, discuss the advantage of using complex valued SAR imagery in ship detection, third present a complex valued robust statistical testing approach suited for ship detection in complex valued SAR where the proposed test uses a complex valued local statistical modelling of the sea clutter and fourth illustrate the performance obtained with the adopted detection approach on complex valued SAR imagery from Sentinel-1 compared to the performance of ship detection methods obtained from Sentinel-1 amplitude focused SAR imagery of similar scene. More details.
Authors/Presenters: Karim Seghouane (University of Melbourne), Cao Tan, Connor Luckett, Sebastien Wong (Defense Science and Technology Group, DSTG) Ritwik Gupta (Defense Innovation Unit and UC Berkeley, USA), and Antonio Robles-kelly (Defense Science and Technology Group, DSTG).