Underwater image enhancement technology has the potential to improve the quality and visibility of underwater images, which has important application value in enriching digital media resources, underwater exploration, underwater communication, and other fields. In recent years, deep learning methods have achieved remarkable results in underwater image enhancement. However, the existing methods suffer from high computational complexity, which limits their application in scenarios with limited computational resources. To address this problem, a lightweight underwater image enhancement method is proposed. The method is based on cross-scale depth distillation feature perception and adopts a U-shaped network structure, which can significantly reduce the model parameter size while ensuring the extraction of nonlinear abstract layers. The experimental results show that the proposed method has achieved competitive results in both visual effects and objective evaluation metrics.