Ship trajectory prediction plays a key role in the early warning and safety of maritime traffic. It is a necessary assistant tool that can forecast a ship’s trajectory in a certain period to prevent ship collision. However, highly precise prediction of long-term ship trajectories is still a challenge. This study proposes a ship trajectory prediction model called ShipTrack-TVAE, which is based on a Variational Autoencoder (SocialVAE) and Transformer architecture. It aims to address ship trajectory prediction tasks in complex waterways. To enable the model to avoid potential collision risks, this study designs a collision avoidance mechanism, which comprehensively incorporates safety constraints related to the distance between ships into the loss function. The experimental results show that on the Qiongzhou Strait ship AIS dataset, the Average Displacement Error (ADE) of ShipTrack-TVAE improved by 21.85% compared to the current state-of-the-art trajectory prediction model, SocialVAE, while the Final Displacement Error (FDE) improved by 17.83%. The experimental results demonstrate that the ShipTrack-TVAE model can effectively improve the prediction accuracy of short-term, medium-term, and long-term ship trajectories. It has excellent performance and provides a certain reference value for advancing unmanned ship collision avoidance.

The safety of ships during nighttime navigation has always been a major concern. With the widespread application of technologies such as intelligent recognition, intelligent detection, and unmanned ship navigation at night, nighttime maritime light pollution has significantly affected the effectiveness of these intelligent technologies and navigation safety. Therefore, effectively eliminating nighttime maritime light pollution has become an urgent challenge that needs to be addressed. This paper presents a model based on spatial frequency blocks (SFBs) to solve the problem of light pollution in nighttime sea images. The model includes ResNet-50, an encoder, a decoder, and a discriminator. To enable the model to better remove the influence of light pollution, this study designs a method of first detecting the light pollution area and then removing it. It extracts image information from the space–frequency domain to help eliminate light pollution and retain more image information. The experimental results show that on the nighttime light pollution dataset, the Peak Signal-to-Noise Ratio (PSNR) of the model is improved to 24.91 compared to the current state-of-the-art image restoration model, while the Frechet inception distance (FID) is reduced to 64.85. At the same time, in the real night environment, the model can better remove light pollution to recover the original nighttime information. It has excellent performance and provides a certain reference value for advancing the safety of nighttime maritime navigation.