Theses and Dissertations

Advisor

Boyd, Dylan

Committee Member

Ball, John

Committee Member

Diao, Junming

Date of Degree

8-12-2026

Original embargo terms

Immediate Worldwide Access

Document Type

Graduate Thesis - Open Access

Major

Electrical & Computer Engineering

Degree Name

Master of Science (M.S.)

College

James Worth Bagley College of Engineering

Department

Department of Electrical and Computer Engineering

Abstract

Maritime Synthetic Aperture Radar (SAR) imagery is frequently degraded by vessel motion and ocean-induced dynamics that violate the stationary scene assumptions of conventional SAR image formation, resulting in azimuth defocus and reduced image interpretability. Autofocus techniques are commonly employed to mitigate these effects; however, their relative performance in maritime environments and the tradeoffs between image quality and computational complexity remain areas of ongoing investigation. This work presents a comparative evaluation of three established SAR autofocus algorithms—Phase Gradient Autofocus (PGA), Minimum Entropy (ME), and the Doppler Parameter Estimation Algorithm (DPEA)—applied to real-world maritime SAR imagery. The study utilizes X-band SAR data from the Capella Space Open Data Library, consisting of both spotlight and stripmap acquisitions containing visibly defocused vessel targets. Algorithm performance is evaluated using a combination of qualitative image assessment and quantitative metrics, including entropy, RMS contrast, peak sidelobe ratio (PSLR), and integrated sidelobe ratio (ISLR). Results from single-algorithm analysis indicate that PGA is effective for correcting mild defocus but exhibits limited capability for more severe motion-induced blur. Both ME and DPEA provide substantial improvements across a wider range of defocus conditions, with ME generally producing stronger performance in radar-based metrics and DPEA achieving more consistent convergence and improving a larger portion of the dataset. Sequential permutation experiments demonstrate that combining algorithms significantly improves performance, though results depend strongly on algorithm ordering. Among the evaluated combinations, PGA-DPEA-ME produces the strongest visual results, while DPEA-PGA-ME achieves the best performance in image-based quantitative metrics. Permutations beginning with ME provide improved sidelobe suppression but at increased computational cost. Additional analysis of reduced-complexity combinations shows that DPEAPGA and PGA-DPEA provide strong performance across both qualitative and quantitative metrics without the inclusion of ME. In many cases, these combinations produce results comparable to full permutations while requiring significantly less processing time.

Download

Share

COinS