Design of Underwater Rescue System Based on Sonar Detection and Polarization Imaging

Underwater detection technology is crucial for marine resource development, emergency rescue, and other fields, but a single sensor has inherent limitations. Developing a multimodal fusion system can break through the physical limitations of a single sensor and improve the detection reliability of complex underwater environments. This study addresses the issues of low target recognition accuracy and short detection distance in low light and high scattering environments, and designs a collaborative detection system that combines sonar detection and polarization imaging. By improving the Jaffe model, quantifying the optimization effect of system spacing on contrast, and designing multi-level image processing algorithms, the resolution of sonar optical images has been increased to 40 × 40 pixels, and the recognition rate has been improved to 65%. The research results indicate that the multimodal system can extend the effective detection range to 15 meters and increase the identification rate by 15% under specific water quality conditions, making it suitable for scenarios such as closed water rescue. The system has verified the engineering feasibility of the sonar optical imaging collaborative model, providing data support for the development of miniaturized devices. However, its performance is limited by water quality transparency and lighting conditions, and further optimization of scattering suppression algorithms is needed to adapt to more complex environments.