Significance based Ship Detection from SAR Imagery

Synthetic Aperture Radar images have potential applications in the surveillance scenario and hence automated target detection algorithms prove to be a useful tool in monitoring and crime control as well as in marine traffic management. The advancements in marine trade have lead to the increase in the number of ships in the world waters. The usage of satellite-based radar images have become well known for maritime surveillance as ship detection is relatively simple and independent of the climatic conditions. Ships can be easily discerned in the SAR images due to their bright intensity which results due to the strong radar backscatter from their metal surface. These are the significant pixels in an image which can be gathered to detect the ship targets. During heavy sea state conditions and presence of speckle noise, sea ice and coastline structure, the ship detection process is affected since these non-ship features in the sea also exhibit high intensities in the SAR image. These false alarms have to be reduced. So, in this work a Significance based ship detection algorithm followed by a discrimination stage using ensemble classifier is proposed to differentiate the ship and non-ship targets. To enhance the ship detection process, the images are subjected to ridgelet transform based despeckling. The efficacy of the proposed Significance based target detection is proved by the obtained results.