Contrastive Learning-Based Haze Visibility Enhancement in Intelligent Maritime Transportation System

With the rapid development of artificial intelligence and big traffic data, data-driven intelligent maritime transportation has received significant attention in both industry academia. It is capable improving efficiency reducing accidents applications. However, video cameras often suffer from severe haze weather, leading to degraded visual data ineffective surveillance. thus necessary restore visually images guarantee safety under hazy imaging conditions. In this work, a contrastive learning framework proposed for visibility enhancement systems. particular, method could fully learn local global image features, which are beneficial quality improvement. A total 100 clean containing water scenes were selected as synthetic test dataset, good dehazing results achieved on indexing (e.g., peak signal noise ratio (PSNR): 23.95 ± 3.48 structural similarity index (SSIM): id="M2"> 0.924 0.065 different transmittance atmospheric light values). addition, extensive experiments real-world demonstrate effectiveness natural evaluator (NIQE): id="M3"> 4.800 0.634 perception-based (PIQE): id="M4"> 46.320 10.253 ). The enhanced be effectively exploited promoting accuracy robustness ship detection. supervision management accordingly improved system.