Wavelet transform vision state space model for underwater image enhancement

Abstract

Underwater image enhancement technology is crucial for improving the visual quality of marine target detection, robot navigation and biological research. However, underwater images often suffer from severe color distortion, reduced contrast and blurred details due to complex optical effects such as water absorption, scattering and uneven illumination. To this end, we propose a wavelet transform vision enhancement framework based on the State Space Model, combined with wavelet transform, to effectively restore structural information and correct color deviation. Amplitude-phase separation and reconstruction are carried out through the dual-branch upsampling mechanism of deep Fourier transform, significantly suppressing scattering noise to restore high-frequency details. Meanwhile, the channel-wise perception fusion strategy is adopted to achieve cross-scale feature interaction and channel adaptive weighting, optimizing the naturalness of color and visual consistency. The experimental results prove the effectiveness of this method on multiple datasets.