基于小波分解与Retinex校正的水下鱼群图像增强系统设计

Design of an Underwater Fish School Image Enhancement System Based on Wavelet Decomposition and Retinex Correction

摘要: 针对水下复杂成像环境中光照衰减严重、颜色偏移明显以及关键细节缺失等技术退化问题，本文设计并实现了一种基于小波分解与Retinex校正的水下鱼群图像增强系统。该系统在预处理阶段首先利用自适应直方图均衡化(AHE)优化低对比度区域；随后通过小波变换对退化图像进行多尺度分层解耦，将其划分为承载宏观轮廓信息的低频分量与包含纹理特征的高频分量。针对低频分量，引入Retinex色彩恒常性模型进行光照估计与色度校正，以消除水体散射带来的色彩失真；针对高频分量，实施细节纹理增益处理。最后，采用多尺度融合策略对差异化增强后的子带进行金字塔重建。在MATLAB平台上的仿真实验结果表明，该系统能显著改善水下鱼群图像的视觉可视度，客观评价指标峰值信噪比(PSNR)与水下图像质量评价指标(UIQM)均有明显提升，具有良好的工程应用价值。

Abstract: To address technical degradation issues in complex underwater imaging environments—including severe light attenuation, significant color distortion, and loss of critical details—an underwater fish school image enhancement system based on wavelet decomposition and Retinex correction was designed and implemented in this paper. During the preprocessing stage, adaptive histogram equalization (AHE) is employed to optimize low-contrast regions. Wavelet transformation then performs multi-scale hierarchical decoupling of degraded images, separating them into low-frequency components containing macroscopic contour information and high-frequency components carrying texture features. For low-frequency components, the Retinex color constancy model is applied for illumination estimation and chromatic correction to eliminate color distortion caused by water scattering; for high-frequency components, detail and texture enhancement techniques are applied. Finally, a multi-scale fusion strategy is utilized to perform pyramid reconstruction on enhanced subbands. Simulation results on the MATLAB platform demonstrate that this system significantly improves the visual clarity of underwater fish school images, with notable improvements observed in both peak signal-to-noise ratio (PSNR) and underwater image quality metrics (UIQM), highlighting its strong practical value for engineering applications.