Ship Target Rapid Detection and Signal Extraction in Wide Area Oceanic Scenes

This research project presents a framework for rapid detection of ship targets and extraction of associated signals from synthetic aperture radar (SAR) range-compressed data covering large oceanic scenes. The conventional processing chain for ship detection requires full SAR image focusing, which introduces substantial computational overhead. To address this limitation, the project utilizes the RCShip dataset, constructed specifically for range-compressed domain analysis. Two detection algorithms are implemented and evaluated: KAN²-DET, which separates target detection from focusing to enable wide-area coverage, and RMY-SAR, which employs multi-layer saliency and contour extraction for accurate target identification in heterogeneous backgrounds. A complete web-based platform is developed to demonstrate the functionality. The front-end uses HTML, CSS, and JavaScript to provide modules including home, registration, login, dashboard, image detection, live detection, and logout. The back-end is built with Python and the Flask framework, while SQLite serves as the embedded database for user management and result logging. The system accepts SAR image inputs, processes them through the selected algorithms, and returns detected ship positions and signal characteristics. This integrated approach reduces processing latency compared to fully focused methods and provides an accessible interface for research validation.

Keywords: Ship detection, SAR range-compressed data, KAN²-DET, RMY-SAR, RCShip dataset, signal extraction, oceanic scenes, Flask framework, SQLite, web platform.