Advancing Maritime Surveillance: Multi-Layer Perceptron Models for Enhanced Vessel Detection from SAR Imagery

This paper presents a novel approach to enhancing maritime surveillance through advanced object detection in Synthetic Aperture Radar (SAR) imagery, focusing on the identification of small vessels. Addressing the challenge of detecting minute objects in high-resolution images, the study introduces a simple yet effective model that integrates traditional image processing techniques with innovative methodologies. Utilizing a Multilayer Perceptron (MLP) model implemented in Python, the research offers a detailed comparison with existing techniques, showcasing improved detection and classification capabilities. The novelty lies in integrating advanced preprocessing methods with sophisticated feature extraction via GLCM, enhancing the MLP's input for more accurate detection. This work promises improvements in computational efficiency without sacrificing accuracy, potentially setting a new benchmark for real-time maritime surveillance systems. This paper includes a comprehensive review of related works, a mathematical framework for the proposed model, and experimental results that demonstrate its effectiveness in various image resolutions and types, contributing significantly to the field of maritime surveillance and SAR image analysis.