Colour transfer between images

• 主要作者: Neoh, Javier Hong Rui
• 其他作者: He Ying
• 格式: Final Year Project
• 語言: English
• 出版: Nanyang Technological University 2025
• 主題: Computer and Information Science; Colour transfer; Computer vision; Semantic segmentation
• 在線閱讀: https://hdl.handle.net/10356/184130
• 機構: Nanyang Technological University

Colour transfer between images is a process aimed at adjusting the colour composition of one image to match the reference colours of another. The goal of this project is to implement and compare two colour transfer algorithms: Reinhard et al.’s method (2001) and Pitié et al.’s method (2005), assessing their performance in terms of image quality and how they preserve details and the overall quality of the colour transfer. This report also explores the potential improvement in colour transfer quality by incorporating semantic segmentation techniques, such as the Segment Anything Model (SAM) and YOLO, to enhance contextual awareness in the process. In both areas, we will explore various test cases and evaluates how each algorithm handles different image compositions and colour transfers. The findings suggest that while both algorithms have their merits, context-aware segmentation significantly improves the results, allowing for more accurate and visually consistent colour transfers. This paper contributes to ongoing research in image processing by proposing an integrated approach combining traditional methods with modern segmentation techniques for enhanced image colour manipulation.