A Novel Parameter Estimation Method of Motion Error in Terahertz SAR Imaging Based on Viterbi and RANSAC Algorithms

Abstract

Terahertz synthetic aperture radar (THz SAR) imaging has gradually become one of the hotspots with the development of THz technology. The wide bandwidth can provide satisfactory resolution in the SAR imaging, whereas the small wavelength makes the imaging system more sensitive to the platform vibration error (VE), which may seriously affect its image quality. In this article, a novel parameters estimation method of motion error in the THz SAR imaging based on the Viterbi algorithm (VA) and random sample consensus (RANSAC) algorithm is proposed. First, the VA is employed to extract the instantaneous frequency (IF) introduced by the platform VE. The VA is defined based on the generalization of IF characteristics, and it further considers the relationship between the frequency points at adjacent time points based on the maximum value search in the time–frequency representation, which makes it perform well in the IF extraction. Second, the RANSAC algorithm in conjunction with the nonlinear least square (NLS) algorithm and the minimum-entropy principle is proposed to perform the VE parameters estimation. Due to the noise or the platform velocity variation, there will exist the points that deviate from the high-frequency VE, which can be mitigated via the combination of the RANSAC algorithm and the NLS technique. Simultaneously, the well-focused THz SAR imaging results can be acquired after compensating via the best parameter estimation results selected by the minimum-entropy principle. Finally, the validity of the proposed method has been demonstrated through the simulation and real-measured experiments.

Keywords

• Platform vibration error (VE) compensation
• random sample consensus (RANSAC) algorithm
• terahertz synthetic aperture radar (THz SAR)
• Viterbi algorithm (VA)