Ship trajectory prediction plays a crucial role in maritime safety and navigation, especially in complex environments where multiple ships interact simultaneously. In maritime applications, exact single trajectory prediction may not be paramount; rather, balancing accuracy and diversity of trajectories is essential to ensure collision avoidance based on predicted paths during ship interactions. However, due to the stochastic nature of ships’ future intentions, resulting in diverse motion patterns and multiple possible paths, predicting trajectories in these regions poses significant challenges. This paper proposes a novel method, Environmental Perception based on Denoising Diffusion Models(EP-DDM), for ship trajectory prediction in multi-ship interaction scenarios, utilizing denoising diffusion models to enhance environmental perception. EP-DDM is based on an encoder-decoder architecture but differs from conventional methods by avoiding the use of latent variables to represent multimodality. The encoder leverages a transformer to extract historical and interactional information, while the decoder, also based on a transformer, adequately captures temporal dependencies. By integrating denoising techniques into the prediction model, this method aims to improve trajectory prediction accuracy while considering trajectory diversity, particularly in scenes involving complex ship interactions. Experimental results on extensive natural ship trajectory data demonstrate that the proposed model achieves excellent prediction performance and robustness compared to baseline methods, highlighting the potential of our approach to enhance maritime trajectory prediction capabilities in dynamic and congested ocean environments.