Underwater Image Enhancement using Double-Stage Multi-Pipeline Preprocessing with Texture–Edge–Contrast–Guided Adaptive Weighted Wavelet Fusion

Abstract Underwater imaging is inherently challenged by wavelength-dependent light absorption and scattering, which cause severe color distortion, low contrast, reduced visibility, and loss of structural information. These degradations significantly affect the visual quality and reliability of underwater images for subsequent analysis and interpretation. This study proposes an adaptive underwater image enhancement framework that uses double-stage multi-pipeline preprocessing and texture–edge–contrast-guided adaptive weighted wavelet fusion to simultaneously improve color fidelity, contrast, visibility, and structural preservation. The proposed framework integrates three complementary preprocessing pipelines. The first pipeline performs gamma-based color correction, followed by double-stage Gaussian filtering to normalize illumination and suppress noise. The second pipeline employs global–local histogram equalization to enhance contrast and expand the perceptual dynamic range. The third pipeline applies image sharpening, then median filtering, to strengthen edge information while preserving fine textures. The outputs of all pipelines are subsequently fused in the wavelet domain using an adaptive local-energy weighting strategy guided by texture, edge, and contrast features, ensuring balanced enhancement across all color channels. Experimentally evaluated the method on the UIEB and SUIM benchmark datasets. The proposed method achieved a Peak Signal-to-Noise Ratio (PSNR) of 33.981 dB, Structural Similarity Index (SSIM) of 0.953, Underwater Color Image Quality Evaluation (UCIQE) of 0.803, and Patch-based Contrast Quality Index (PCQI) of 1.122, outperforming several state-of-the-art underwater image enhancement techniques. Furthermore, increased edge density and richer SIFT features demonstrate preserved structural information and improved feature quality. The proposed framework provides a computationally efficient and structurally reliable solution for underwater image enhancement. By combining multi-pipeline preprocessing with texture–edge–contrast-guided adaptive weighted wavelet fusion, the method effectively enhances color correction, contrast, and detail preservation, resulting in improved visual quality and feature-extraction performance for underwater imaging applications.