ISAR images tracking for extended small vehicles using cubature Kalman MB filter

This work provides a new approach for extended drone targets (EDTs) tracking from inverse synthetic aperture radar (ISAR) images based on robust extended cubature Kalman multi-Bernoulli track-before detect (ECK-MB-TBD) algorithm that handling of the non-linear ISAR observations with Doppler effect. During this work, we present the drone's extension as multiple ellipses with a time-varying orientation angle and solve the resulting inference problem, which involves data association between the nonlinear measurements and sub-ellipses or sub-random matrices model (RMM). In addition, we give a new implementation of the proposed robust ECK-Sub-RMM-MB-TBD filter to estimate the change in the nonlinear orientation of target's shape based on ECK-Beta Gaussian inverse Wishart. Tracking performance is compared to state-of-the-art algorithms on EDTs such as sequential Monte Carlo (SMC)Sub-RMM-MB-TBD and histogram probabilistic multi-hypothesis (H-PMHT)-TBD. The simulation results confirm the effectiveness and robustness of the proposed algorithm. From the results, the SMC-SubRMM-MB-TBD algorithm take significantly more time to compute than the proposed algorithm, because the first algorithm needs a large number of the particles for EDTs. The average running times were 88.63 s for SMC-Sub-RMM-MB-TBD, 38.54 s for ECK-Sub-RMM-MB-TBD and 55.34 s for H-PMHT-TBD. (c) 2022 Elsevier Masson SAS. All rights reserved.