Abstract
Multi-view image processing is pivotal in applications ranging from 3D reconstruction and virtual reality to autonomous driving and medical imaging. Traditional signal models often fall short in capturing the high-dimensional correlation between multiple views. This paper explores the integration of advanced signal models—such as tensor-based decomposition, graph signal processing, and sparse coding—for efficient representation, enhancement, and interpretation of multi-view image data. We examine state-of-the-art algorithms for disparity estimation, depth fusion, and consistent image synthesis across viewpoints. These advanced methods offer a scalable and robust framework for handling redundant and complementary information across views, yielding improved accuracy, compression efficiency, and robustness to occlusions and noise. The proposed models are evaluated across real-world datasets, showing significant performance gains in peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and depth consistency.

This work is licensed under a Creative Commons Attribution 4.0 International License.
Copyright (c) 2024 Dr. Javier Morales (Author)