Fast communicationRange-spread target detection using 2D non-local nonlinear shrinkage map
Introduction
High resolution radar (HRR) can resolve individual scatterers over the radial range extent of a target [1], [2], [3], [4]. Range-spread target detection is attracting more attention from radar community for several reasons. First, HRR provides more details of targets than low resolution radar (LRR). Second, small-size resolution cells reduce the clutter level. Third, radar cross section (RCS) of radar in HRR has much less fluctuation than that in LRR. These facts show that HRR has potential to upgrade target detection capability [5], [6], [7].
In the last decade, many works have been done in range-spread target detection. In [8], the optimal detector based on an ideal assumption on target responses was presented. In [9], generalized likelihood ratio test was developed based on the spatial scattering density (SSD-GLRT). Recently, two detectors in white Gaussian noise were developed to detect range-spread maneuvering targets [6], [7]. In the two detectors, high-resolution range profiles from multiple successive pulses were stacked into a 2D grayscale range-pulse image (RPI). One uses a single-point nonlinear shrinkage map (SPNSM) operating on each high-resolution range profile (HRRP) for noise reduction [6]. The other uses the 2D nonlinear shrinkage map (2DNSM) relevant to the local statistics of the RPI [7]. Moreover, the 2DNSM has better denoising performance than SPNSM [7]. In fact, the nonlinear map for noise reduction plays a key role in range-spread target detection in white Gaussian noise.
In this paper, a novel 2D non-local nonlinear shrinkage map (NLNSM) is proposed for noise reduction and target echoes preservation in the RPI. Replacing the 2DNSM in the GAI-2DNSM of Xu and Shui [7] by the proposed NLNSM, a new detector is developed to detect range-spread target in white Gaussian noise. Experimental results based on measured target data show that the new detector using NLNSM attains better detection performance than the GAI-2DNSM detector in [7].
This paper is organized as follows. Section 2 gives the formulation of the detection problem and the characteristics of range-pulse images. Section 3 presents the NLNSM and the detector that combines the NLNSM with geometric average integration (GAI). Experimental results and performance comparisons are given in Section 4.
Section snippets
Detection problem description
HRR transmits wideband waveforms, such as a direct short pulse, a digital phase coding pulse, a discrete frequency coding pulse and a linear frequency modulated pulse. The target echo of a direct short pulse is a function of time delay or radial range, which is referred to as a complex HRRP. For other types of transmitting pulses, HRRPs are obtained by pulse compression of the received echo [3]. The amplitude of a complex HRRP is named as real HRRP.
Detection problem of range-spread targets
The proposed detector
As shown in Fig. 1(b) and (d), target returns have two obvious characteristics. First, target returns occupy only a small part of the RPI and congregate on a stripe-shaped region. Because of this characteristic, the 2DNSM can efficiently reduce the noise and preserve target echoes in RPIs [7]. Second, target HRRPs vary with pulses due to range walks across cells. With the increase of the interval between two pulses, the corresponding HRRPs suffer from serious range walks and degradation of
Experimental results and performance comparison
Raw radar data in Fig. 1(a) and (c) are used to test the performance of the proposed detector based on the NLNSM. The HRRPs of An-26 and Cessna Citation S/II are regarded as noiseless and they are normalized so that the average power of all the HRRPs equals to one [7]. In this way, detection performance is related to the average RCS of the target during maneuvering flight rather than the RCS at each observation time instant. For a given SNR, noisy real HRRPs are generated by
Conclusion
A new NLNSM method is proposed to improve noise reduction and target echo preservation performance for range-spread target detection. The NLNSM method is heuristic and based on the characteristics of target HRRPs extracted from measured target radar data. Embedding it into the GAI-based detector and EI-based detector, new detectors for maneuvering range-spread targets are obtained. The experimental results using raw radar data of the target show that the detectors using the NLNSM are practical
Acknowledgment
This work was supported by National Natural Science Foundations of China (61201296, 10990012, 61271024 and 61372136), the Fundamental Research Funds for the Central Universities (K5051202037).
References (12)
- et al.
AR-model-based adaptive detection of range-spread targets in compound Gaussian clutter
Signal Process.
(2011) - et al.
Subspace detection for range and Doppler distributed targets with Rao and Wald tests
Signal Process.
(2011) - et al.
An adaptive weighted rank order detector for spatially distributed target
Signal Process.
(2012) High Resolution Radar
(1995)- et al.
Adaptive detection of wideband radar range spread targets with range walking in clutter
IEEE Trans. Aerosp. Electron. Syst.
(2012) - et al.
Range-spread target detection using consecutive HRRPs
IEEE Trans. Aerosp. Electron. Syst.
(2011)
Cited by (9)
A novel detector for range-spread target detection based on HRRP-pursuing
2024, Measurement: Journal of the International Measurement ConfederationAdaptive range-spread maneuveringtarget detection incompound-Gaussian clutter
2015, Digital Signal Processing: A Review JournalCitation Excerpt :Only utilizing energy of target returns limited their performances. In order to utilize abundant detail information of target high resolution range profiles (HRRPs), the feature-based detector using two complex HRRPs [15], the detector using the modified correlation matrix of multiple consecutive HRRPs (MCOM detector) [16], the detectors using two-dimensional (2D) nonlinear shrinkage map [17] and 2D non-local nonlinear shrinkage map [18] were recently developed in our previous works. Owing to the utilization of additional information except energy, the four detectors have good detection performances for range-spread target in white Gaussian noise.
Doppler-Spread Targets Detection for FMCW Radar Using Concurrent RDMs
2022, IEEE Transactions on Vehicular TechnologyRange-spread target detector via coherent energy accumulation and block thresholding denoising
2021, Journal of Systems Engineering and ElectronicsRFD-Rao and RFD-Wald tests for distributed targets with range walking effect
2018, Journal of Central South UniversityManeuvering range-spread target detection in white Gaussian noise using multiple-pulse combined waveform contrast
2017, 2017 IEEE International Conference on Signal Processing, Communications and Computing, ICSPCC 2017