Drone-borne SAR imaging and change detection

Bekar, Ali ORCID: 0000-0002-7773-5529 (2023). Drone-borne SAR imaging and change detection. University of Birmingham. Ph.D.

Preview

Bekar2023PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (10MB) | Preview

Abstract

This thesis aims to carry out fundamental and systematic research on the formation of high-frequency, high-resolution drone-borne Synthetic Aperture Radar (SAR) imagery of an extended target area and the detection of both sizeable object displacements and subtle changes in it. Drone-borne systems are affordable, rapidly deployable, and capable of accessing difficult areas where human access is dangerous or unsafe. However, they are better suited to relatively short-range applications due to their limited payload capacities which restrict transmit powers and sensor sizes. At high operating frequencies, the radar form factor is reduced and the potential for obtaining very fine-resolution imagery and the sensitivity to temporal changes in the scene are improved. On the other hand, short-range operation at high frequencies leads to significant space-variant/invariant errors induced by motion errors resulting from wind turbulence and drone vibration. Further, for temporal change detection that relies on comparison of “before/after” radar imagery, motion errors are unique for each drone pass. Even if two focused images are formed, residual co-registration errors and residual spatially invariant/variant phase errors can degrade the resulting change map.

The research presented in this thesis sets out to develop image formation and change detection algorithms capable of handling the mentioned difficulties above based on theoretical analyses, validate the algorithms using simulations and evaluate the experimental performance in real-world conditions using a drone-borne SAR demonstrator built within this thesis' scope. In the experiments, low-cost vehicular radars operating at 24 and 77 GHz are used, and novel short-range, fine-resolution imagery (up to 1.7 cm in cross-range) of an extended target area is formed without employing a dedicated positional system. Also, it is shown, for the first time, that a high-frequency, high-resolution drone-borne SAR system operating at short ranges is capable of discerning both sizeable object displacements such as human replacement, and subtle changes such as the human footprint or car tyre marks.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):