Research on Machine Learning-Based Underwater Image Enhancement Methods under Multi-Module Fusion Modeling

This study aims to deal with the image degradation of underwater images due to complex optical characteristics and propose a machine learning-based underwater image enhancement method under a multi-module fusion model. The study integrates traditional models such as image statistical analysis model and underwater image degradation model and incorporates machine learning algorithms such as convolutional neural network (e.g., VGG16) and support vector machine (SVM). The depth features of the image are extracted by VGG16 and fused with the traditional manual features, and the SVM is used to realize more accurate image degradation classification. Then, a deep learning image quality assessment model is constructed and combined with traditional metrics such as PSNR, UCIQE and UIQM to comprehensively assess the enhancement effect. This method not only utilizes the multi-module fusion strategy to comprehensively deal with multiple degradation scenarios, but also improves the accuracy and intelligence of the classification and evaluation with the help of machine learning algorithms and enhances the adaptability of the model to complex scenes. The results provide an effective reference for underwater vision system design.