Efficient Onboard SAR Image Processing for Real- Time Ship Detection and Data Reduction

Maritime Authorities face significant challenges in monitoring vast maritime domains and enforcing regulations within their Exclusive Economic Zones (EEZs). Synthetic Aperture Radar (SAR) imaging of Fers an effective means of monitoring maritime activities, including ship traffic, illegal fishing, and piracy. However, SAR images tend to be very large in size, while limited communication bandwidth from aerial vehicles or satellites to ground stations constrains the transmission of high-resolution SAR images. This paper proposes onboard image processing as a solution to overcome these limitations. By processing SAR images directly on the UAV or satellite, we can significantly reduce the amount of raw data transmitted. This enables the transmission of high-resolution images across wider geographic areas while preserving a wider dynamic range for capturing finer details in the scene. However, onboard processing presents its own challenge, namely, limited power availability on satellites and UAV s. To address this, we explore the use of low-power, commercially available of F-the-shelf (COTS) mobile phone processors for onboard SAR image processing. These processors offer a trifecta of benefits: low power consumption, large address space, and compatibility with low-power memory, making them ideal for large-scale image processing onboard. By performing image processing onboard, we optimize data trans-mission, minimize bandwidth usage, and enhance the efficiency of maritime surveillance operations. Our experimental results, using GDAL for image compression and OpenCV for image processing on a Qualcomm Snapdragon Gen 2 mobile processor running Kali Linux, demonstrate a significant improvement in image transmission and storage efficiency, achieving a substantial bandwidth reduction of 91.21 % while maintaining real-time performance and preserving accuracy. Thus our solution paves the way for automated ship detection, tracking, and illegal behavior identification.