CCAC Theses and Dissertations

Campus Access Only

All rights reserved. This publication is intended for use solely by faculty, students, and staff of Nova Southeastern University. No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, now known or later developed, including but not limited to photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author or the publisher.

Date of Award

2026

Document Type

Dissertation - NSU Access Only

Degree Name

Doctor of Philosophy (PhD)

Department

College of Computing, AI, and Cybersecurity

Advisor

Michael Laszlo

Committee Member

Francisco Mitropoulos

Committee Member

Sumitra Mukherjee

Keywords

dual annealing, global optimization, image quilting, minimum error boundary cut, texture synthesis, tiling error

Abstract

Real-time texture synthesis remains challenging due to the tradeoff between computational efficiency and perceptual quality. Traditional patch-based methods, such as Image Quilting with Minimum Error Boundary Cut (MEBC), achieve high visual fidelity but incur significant computational cost and rely on local optimization. This dissertation presents a heuristic-driven framework that repurposes MEBC-derived seam error metrics as reusable signals for global optimization. A modular synthesis system was implemented in Python within the Google Colaboratory environment, integrating seam cost memoization, adaptive blending, and a refinement process using dual annealing. The framework was evaluated using textures from the USC-SIPI Image Database under varying synthesis configurations. Performance was measured using Structural Similarity Index (SSIM), Mean Squared Error (MSE), Summed Overlap Quality (SOQ), Maximum Overlap Quality (MOQ), and runtime, along with qualitative visual analysis. Results demonstrate that the proposed method improves structural coherence and reduces visible seam artifacts compared to baseline approaches, achieving measurable gains in perceptual quality without exhaustive seam recomputation. This work introduces a scalable, seam-aware optimization approach that advances the balance between efficiency and visual fidelity in texture synthesis.

Download
To access this thesis/dissertation you must have a valid nova.edu OR mynsu.nova.edu email address and create an account for NSUWorks.

Free My Thesis

If you are the author of this work and would like to grant permission to make it openly accessible to all, please click the Free My Thesis button.

  Contact Author

Share

COinS