Transformer-based methods have demonstrated remarkable performance in underwater image enhancement due to their ability to capture long-range dependencies, crucial for high-quality reconstruction of degraded images. However, existing Transformer-based techniques often treat all token similarities equally during self-attention, which can lead to the aggregation of irrelevant features, hampering clear image restoration. We propose U-ENHANCE, a novel Underwater image Enhancement framework that integrates wavelet-based frequency decomposition with spatial domain attention to address these challenges. In particular, we introduce a Wavelet Triple Self-Attention (WTSA) mechanism that performs self-attention across three dimensions—horizontal, vertical, and channel-wise, effectively capturing multi-scale features critical for restoring fine details and structural integrity. Additionally, we design a Self-Calibrated Feedforward Network (SCFN) that refines feature representation by dynamically adjusting the receptive field, further enhancing spatial and frequency domain integration. Extensive experiments on underwater image enhancement benchmarks demonstrate that U-ENHANCE outperforms state-of-the-art methods by providing superior restoration of color, clarity, and structural details. The code is available at: https://github.com/Priyanka01mishra/UENHANCE .