Use this URL to cite or link to this record in EThOS:
Title: Phase correlation-based illumination-insensitive image matching for terrain-related applications
Author: Wan, Xue
ISNI:       0000 0004 7233 0826
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
The Earth surface is monitored by air-borne and space-borne sensors locally, regionally and globally and at short time intervals. The capability of multi-temporal image matching is essential for many terrain-related applications, such as glacier motion monitoring, landslide localisation and earthquake deformation assessment. Robust image matching using time-varying remotely sensed data is, however, not always achievable because images are acquired separately, under different illumination conditions. For remotely sensed images of natural landscapes, such as mountains, deserts and glaciers, the majority of image patterns is produced by topographic features, such as shading and shadows. The appearances of these image patterns can be greatly altered by illumination variation, especially local illumination variation caused by solar position change, and thus parts of the images may lose correspondence for accurate matching. This research provides a thorough mathematical analysis of the impact of local illumination variation via Phase Correlation (PC) cross power spectrum in the Fourier frequency domain and thus proves the illumination-insensitive property of PC-based image matching algorithms. Specifically, the relationship between PC fringe appearance and sun angle variation is investigated. Two improved PC based approaches, ADCF-PC (Absolute Dirichlet Curve Fitting based Phase Correlation) and PLSF-PC (Piecewise Least Square Fitting based Phase Correlation), are presented and compared for illumination-insensitive image matching. Finally, three original approaches to explore terrain related applications have been developed including: vision-based UAV navigation, illumination-insensitive change detection and DEM generation from multi-temporal remotely sensed images.
Supervisor: Liu, Jianguo Sponsor: Imperial College London
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral