Abstract
This paper analyzes sea clutter by a random series without assuming the scattering being independent. We quantitated the complexity of sea clutter by applying multiscale sample entropy. We found that above certain wave heights or wind speeds, and for HH or VV polarization, the target can be distinguished from sea clutter by regarding (i) the sample entropy at large scale factors or (ii) the complexity index (CI) as entropy metrics. This is because the backscattering amplitudes of range bins with the primary target were found equipped with the lowest sample entropy at large scale factors or the lowest CI compared to that of range bins with sea clutter only. To further cover low-to-moderate sea states, we constructed a polarized complexity index (PCI) based on the polarization signatures of the multiscale sample entropy of sea clutter. We demonstrated that the PCI is yet another alternative entropy metric and can achieve a superb performance on distinguishing targets within 1993’s IPIX radar data sets. In each data set, the range bins with the primary target turned to have the lowest PCI compared to that of range bins with sea clutter alone. Moreover, in our experiment using 1993’s IPIX radar data sets, the PCIs of range bins with sea clutter only were almost the same and stable in each data set, further suggesting that the proposed PCI metric can be applied in the presence of no or multiple targets through proper fitting curves.
Highlights
Detecting surface targets on the sea surface finds wide applications, including but not limited to navigation security, law enforcement, maritime surveillance, and emergency responses or rescues [1], just to name a few
The rest of this paper shows our endeavors in detail and is organized as follows: Section 2 first introduces the data sets we selected as test samples, Section 3 presents the details and discussions of the multiscale sample entropy (MSE) algorithm and how the proposed entropy metrics were constructed step by step, Section 4 validates and discusses the performances of the proposed entropy metrics, and a conclusion is drawn to close this paper
* The value of the range bin with the primary target is highlighted in both italic and bold for each data set. Together with these polarization signatures of the MSE, we sought a nonlinear relationship that (i) keeps the dominance of the HH- or VV-polarized MSE under high sea conditions to ensure that the target is detectable as the two former metrics did under high sea conditions, and (ii) compensates the MSE of range bins with pure sea clutter more but the MSE of range bins with targets less, to ensure that the targets are distinguishable by their lowest complexity under low or moderate sea conditions
Summary
Detecting surface targets on the sea surface finds wide applications, including but not limited to navigation security (e.g., anti-collisions), law enforcement (e.g., illegal fishing), maritime surveillance, and emergency responses or rescues [1], just to name a few. 2021, 13, 3950 range bins selected from the data sets (for more details, see Section 2) we applied in this study, accompanied by their estimated shape parameters, which are dependent on the range bin size, incident grazing angle, sea states, and radar parameters (e.g., operating frequency, resolutions) It has been observed that a significant difference in the estimated Hurst parameter exists between the range bins with and without a target under sea clutter, implying that the Hurst parameter can be an alternative metric for detecting surface targets (HH polarization only). The rest of this paper shows our endeavors in detail and is organized as follows: Section 2 first introduces the data sets we selected as test samples, Section 3 presents the details and discussions of the MSE algorithm and how the proposed entropy metrics were constructed step by step, Section 4 validates and discusses the performances of the proposed entropy metrics, and a conclusion is drawn to close this paper
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call