Author

Yutong Ge

Date of Award

6-9-2024

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2024

Degree Name

Master of Science (MS)

Department

Computer Science and Engineering

First Advisor

Nam Ling

Abstract

In the field of image reconstruction and super-resolution, using codebooks has shown promising results despite various image degradations. Previous methods either use distinct codebooks for each image category or multiple codebooks per category, with the latter achieving better performance by capturing more nuanced image features. Our research proposes a novel method that employs enhanced sets of codebooks and weight maps tailored to each image category. These weight maps dynamically combine different codebook bases to adapt to various reconstruction tasks, resulting in improved image recognition and robustness. This approach significantly enhances the expressiveness and quality of reconstructed images, making it versatile and effective for handling diverse image degradation scenarios.

Share

COinS