Saliency modelling for perceptual image processing

Zhao, Xin 2023. Saliency modelling for perceptual image processing. PhD Thesis, Cardiff University. Item availability restricted.

	PDF (Xin Zhao, PhD, Thesis) - Accepted Post-Print Version Restricted to Repository staff only until 27 March 2026 due to copyright restrictions. Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (36MB)
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (292kB)

Abstract

In the field of perceptual image processing, our research emphasises the critical role of saliency modelling, particularly in addressing distortion-induced saliency variation (DSV). We have developed the MDSV metric, which utilises convex optimisation to minimise the error between the MDSV metric and the Difference Mean Saliency Variation Score (DMSS) by dynamically integrating local and global saliency measures. This metric significantly enhances the accuracy of DSV quantification, crucial for image quality assessment. Complementing this, we introduce the CUDAS database, a pioneering benchmark for distortion-aware saliency, encompassing diverse eye-tracking data. CUDAS provides a comprehensive platform for evaluating and refining saliency prediction models, addressing the challenges posed by diverse natural scenes and distorted images. Moreover, to align computational models with human perception in saliency mapping, we conducted targeted subjective experiments. These led to the development of the perception-based visual saliency evaluation metric (P-VSEM), bridging the gap between technical metrics and human perception. P-VSEM not only advances our understanding of saliency but also aids in evolving fixation prediction models for perceptual image processing. This thesis will capture the essence of the research in saliency modelling, its application in perceptual image processing, and the integration of human perception studies.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Schools > Computer Science & Informatics
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Date of First Compliant Deposit:	27 March 2025
Date of Acceptance:	7 March 2025
Last Modified:	28 Mar 2025 14:27
URI:	https://orca.cardiff.ac.uk/id/eprint/176836

Actions (repository staff only)

Edit Item